Hemisferio Derecho Comprension

doi:10.
1093/brain/awy270 BRAIN 2018: 141; 3389–3404 | 3389
How right hemisphere damage after stroke

can impair speech comprehension
Andrea Gajardo-Vidal,1,2 Diego L. Lorca-Puls,1 Thomas M. H. Hope,1 Oiwi Parker Jones,3
Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Mohamed L. Seghier,1,4 Susan Prejawa,5 Jennifer T. Crinion,6 Alex P. Leff,6,7
David W. Green8 and Cathy J. Price1
See Sheppard and Hillis (doi:10.1093/brain/awy291) for a scientific commentary on this article.
Acquired language disorders after stroke are strongly associated with left hemisphere damage. When language difficulties are observed
in the context of right hemisphere strokes, patients are usually considered to have atypical functional anatomy. By systematically
integrating behavioural and lesion data from brain damaged patients with functional MRI data from neurologically normal partici-
pants, we investigated when and why right hemisphere strokes cause language disorders. Experiment 1 studied right-handed patients
with unilateral strokes that damaged the right (n = 109) or left (n = 369) hemispheres. The most frequently impaired language task
was: auditory sentence-to-picture matching after right hemisphere strokes; and spoken picture description after left hemisphere
strokes. For those with auditory sentence-to-picture matching impairments after right hemisphere strokes, the majority (n = 9) had
normal performance on tests of perceptual (visual or auditory) and linguistic (semantic, phonological or syntactic) processing.
Experiment 2 found that these nine patients had significantly more damage to dorsal parts of the superior longitudinal fasciculus
and the right inferior frontal sulcus compared to 75 other patients who also had right hemisphere strokes but were not impaired on
the auditory sentence-to-picture matching task. Damage to these right hemisphere regions caused long-term speech comprehension
difficulties in 67% of patients. Experiments 3 and 4 used functional MRI in two groups of 25 neurologically normal individuals to
show that within the regions identified by Experiment 2, the right inferior frontal sulcus was normally activated by (i) auditory
sentence-to-picture matching; and (ii) one-back matching when the demands on linguistic and non-linguistic working memory were
high. Together, these experiments demonstrate that the right inferior frontal cortex contributes to linguistic and non-linguistic
working memory capacity (executive function) that is needed for normal speech comprehension. Our results link previously unrelated
literatures on the role of the right inferior frontal cortex in executive processing and the role of executive processing in sentence
comprehension; which in turn helps to explain why right inferior frontal activity has previously been reported to increase during
recovery of language function after left hemisphere stroke. The clinical relevance of our findings is that the detrimental effect of right
hemisphere strokes on language is (i) much greater than expected; (ii) frequently observed after damage to the right inferior frontal
sulcus; (iii) task dependent; (iv) different to the type of impairments observed after left hemisphere strokes; and (v) can result in long-
lasting deficits that are (vi) not the consequence of atypical language lateralization.
1 Wellcome Centre for Human Neuroimaging, UCL Queen Square Institute of Neurology, London WC1N 3AR, UK
2 Faculty of Health Sciences, Universidad del Desarrollo, Concepcion 4070001, Chile
3 FMRIB Centre, University of Oxford, Oxford OX3 9DU, UK
4 Cognitive Neuroimaging Unit, Emirates College for Advanced Education, PO Box 126662, Abu Dhabi, UAE
5 Department of Neurology, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig 04103, Germany
6 Institute of Cognitive Neuroscience, University College London, London WC1N 3AR, UK
7 Department of Brain Repair and Rehabilitation, UCL Queen Square Institute of Neurology, London WC1N 3AR, UK
8 Experimental Psychology, Faculty of Brain Sciences, University College London, London WC1H 0AP, UK
Correspondence to: Andrea Gajardo-Vidal

Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
Received February 1, 2018. Revised August 16, 2018. Accepted September 12, 2018. Advance Access publication November 9, 2018
ß The Author(s) (2018). Published by Oxford University Press on behalf of the Guarantors of Brain.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse,
distribution, and reproduction in any medium, provided the original work is properly cited.
3390 | BRAIN 2018: 141; 3389–3404 A. Gajardo-Vidal et al.
12 Queen Square
WC1N 3AR, London, UK
E-mail: [email protected]
Keywords: right-hemisphere stroke; lesion-deficit mapping; functional MRI; sentence comprehension; working memory
Abbreviation: VBM = voxel-based morphometry
expect the regions that are damaged in the language im-

Introduction paired patients to be activated when neurologically normal
Despite a long history of research favouring the view that participants are performing the affected language tasks.
the left hemisphere is dominant for language processing in Conversely, if impairments on language tasks after right
most right-handed subjects, there is accumulating evidence hemisphere damage are always the consequence of atypical

that the right hemisphere contributes to (i) language func- premorbid language processing in the right hemisphere then
tion in neurologically normal individuals (Hartwigsen we would not expect the regions that are damaged in lan-
et al., 2010a, b; Sollmann et al., 2014); and (ii) language guage impaired patients to be activated when right-handed
recovery after (a) left-hemisphere brain damage (Crinion neurologically normal participants are processing linguistic
and Price, 2005; Thiel et al., 2006; Forkel et al., 2014; material. We would, however, expect that if patients have
Xing et al., 2016; Nardo et al., 2017); or (b) disruption atypical language lateralization (e.g. at a semantic, phono-
of left-hemisphere processing (Hartwigsen et al., 2013; Jung logical or syntactic level), then they would have impair-
and Lambon Ralph, 2016). For example, studies of the ments on all the tasks that tapped the affected language
neurologically normal brain suggest that bilateral inferior function. The evidence sought therefore rests on a combin-
frontal and insula regions support the mapping from sound ation of: (i) neuropsychological assessments of patients with
to lexical meaning (Bozic et al., 2010), bilateral anterior right hemisphere damage to ascertain what level of process-
temporal lobes are involved in the representation of con- ing is affected in those who have impaired performance on
ceptual knowledge (Rice et al., 2015; Jung and Lambon language tasks; (ii) lesion analyses to identify which right
Ralph, 2016; Lambon Ralph et al., 2017) and bilateral hemisphere brain regions were required pre-morbidly for
inferior frontal and supramarginal gyri contribute to the language tasks the patients had difficulty with; and
phonological decisions (Hartwigsen et al., 2010a, b) and (iii) functional neuroimaging studies of neurologically
speech production (Sollmann et al., 2014). For some lan- normal individuals to determine whether the right hemi-
guage tasks, right-hemisphere activation may be driven by sphere regions damaged in the patients with language im-
non-linguistic perceptual processing (Baumgaertner et al., pairments are normally involved in language tasks; and
2013), or the recruitment of attention and working whether their function is more consistent with linguistic
memory (Vigneau et al., 2011). It has also been argued or non-linguistic processing.
that increased right inferior frontal activation in patients Rather than investigate all the different ways that lan-
recovering from aphasia after left hemisphere stroke may guage can break down after right hemisphere damage, the
be the result of upregulating non-linguistic cognitive pro- goal of Experiment 1 was to investigate the language pro-
cessing (van Oers et al., 2010) and the control of semantic cessing impairment that was most frequently observed after
retrieval (Thompson et al., 2016, 2018). right hemisphere strokes. This involved (i) identifying
The hypothesis that right hemisphere executive process- which language task from our standard assessment battery
ing is necessary for normal language function contrasts was most frequently impaired in 109 stroke survivors
with the dominant view that when language impairments (all right-handed and native English speakers) who had
are observed in right-handed patients with unilateral right unilateral right hemisphere damage; (ii) selecting a sample
hemisphere damage, they necessarily imply atypical lan- of patients that were impaired on this language task in the
guage lateralization prior to the stroke (Marien et al., context of spared performance on other tasks (e.g. they
2004). However, the executive processing hypothesis is were all able to recognize objects and match the semantic
consistent with other literature that has associated right content of pictures); and (iii) considering what underlying
hemisphere activation with selective attention (Corbetta processing deficit could explain why the selected patients
et al., 2005; Hillis et al., 2005; Bartolomeo et al., 2012), were impaired on the identified language task. By including
imagery and domain general inhibitory control mechanisms data from 369 right-handed patients with unilateral left
(Aron et al., 2004, 2014; Neef et al., 2018) and working hemisphere strokes, we were also able to compare the lan-
memory (Jonides et al., 1993; Ravizza et al., 2005). guage task that was most frequently impaired after right
Here we investigated whether a subset of impairments on hemisphere strokes to the language task that was most fre-
language tasks after right hemisphere stroke in right- quently impaired after left hemisphere strokes.
handed subjects can be explained by disruption to functions In Experiment 2, we investigated: (i) which right hemi-
that are normally required for language. If so, we would sphere regions were most frequently damaged in those
Speech comprehension after right-hemisphere stroke BRAIN 2018: 141; 3389–3404 | 3391
patients with impaired performance on the language task hemisphere damage; (iii) native speakers of English; (iv) right-
identified in Experiment 1; and (ii) how frequently damage handed prior to the stroke onset; and (v) tested 43 months
to these regions was observed in other patients who had and 510 years after their stroke. Our selection criteria did not
right hemisphere lesions that did not impair performance consider the within hemisphere site of the lesion or the presence
or absence of aphasia. These criteria were met by 109 patients
on the language task identified in Experiment 1. This
with right hemisphere strokes and 369 patients with left hemi-
allowed us to establish whether the identified lesion sites
sphere strokes (see Tables 1 and 2 for demographic, clinical and
were common or rare; and whether the effect of the lesion behavioural data). For each of these groups, we identified and
sites was typical or atypical. compared the language task that was most frequently impaired.
In Experiment 3, we used functional MRI to define which For the right hemisphere patients, we also investigated how
parts of the regions identified in Experiment 2, if any, were consistently impairments on the most frequently affected task
activated when neurologically normal participants per- co-occurred with impairments on other tasks. This allowed us
formed similar tasks to those used in Experiment 1. This to generate hypotheses about the level of processing that was
allowed us to (i) pinpoint which parts of the identified most likely to be affected; and to identify a subset of patients

lesion sites were actively involved in the most frequently with consistent neuropsychological profiles that could be investi-
impaired task; and (ii) reveal how these areas contribute gated further with lesion analyses.
to this task by studying how they respond in other lan-
guage tasks. Finally, in Experiment 4, we investigated the Experiment 2: Which regions are
function of the identified regions further by reporting a damaged in the patients of interest
second functional MRI study of neurologically normal par-
ticipants that examined how activation varied over a range from Experiment 1?
of conditions that differed in their demands on linguistic First, we used voxel-based morphometry (VBM; Ashburner
and non-linguistic working memory. and Friston, 2000), implemented in SPM12 with an unequal
By systematically integrating data from behavioural, variance two sample t-test (Mummery et al., 2000), to identify
lesion and functional imaging studies, we localize right which regions were significantly more damaged in our patients
hemisphere brain structures that support normal language; of interest than in other (control) patients with right hemi-
evaluate their linguistic or non-linguistic functions and offer sphere strokes that did not affect performance on the task
identified in Experiment 1. Details of these patient groups
an explanation for how right hemisphere stroke damage
are provided in the ‘Results’ section. Second, we individually
can impair language performance. examined the lesion site in each patient with the impairment of
interest to establish (i) how many had damage to the regions
identified by VBM; and (ii) which regions were damaged in
Materials and methods those who preserved the lesion sites identified by VBM. Third,
in all other patients, we examined how frequently each of the
This study was approved by the London Queen Square regions of interest was damaged in the absence of an impair-
Research Ethics Committee. ment on the task identified in Experiment 1.
As in Price et al. (2010), Gajardo-Vidal et al. (2018) and
Lorca-Puls et al. (2018), the lesion images entered into the
Experiment 1: Which language task is VBM analysis were continuous measurements of structural
most frequently impaired after right ‘abnormality’ at each and every voxel, calculated by compar-
ing estimates of grey and white matter in each patient to a
hemisphere stroke? sample of 64 neurologically normal controls (Seghier et al.,
Data were extracted from the PLORAS database, which 2008). The output is a 3D lesion image in standard MNI
holds the results of language assessments and high resolution space, indicating the degree of structural abnormality on a
T1-weighted structural MRI brain scans from hundreds of continuous scale from 0 (completely normal) to 1 (completely
stroke survivors recruited in the UK, months to years after abnormal) with a voxel size of 2 2 2 mm3. The advantages
their stroke (Seghier et al., 2016). Language abilities in all of using the fuzzy lesion images have been described in full
patients were assessed using the Comprehensive Aphasia Test elsewhere (Price et al., 2010; Gajardo-Vidal et al., 2018).
(Swinburn et al., 2004), which uses standardized procedures, The analysis included lesion size as a covariate of no interest
and T-scores, to classify each patient’s performance as normal and the search volume was limited to voxels that were classified
or impaired on 27 different cognitive and language tasks as lesioned in at least five patients (as in Fridriksson et al., 2016;
(Table 1). Brain scans were high resolution T1-weighted MRI for rationale, see Sperber and Karnath, 2017). Given that large
acquired with 176 sagittal slices and a matrix size of lesions are more likely to damage critical regions, the inclusion
256 224, yielding a final spatial resolution of 1 mm isotropic of lesion volume as a covariate of no interest in voxel-based
voxels. analyses might have a negative impact on the identification
Patients for the current study were selected according to the of significant lesion-deficit associations (for more details, see
following inclusion criteria: (i) unilateral stroke attested by a clin- Butler et al., 2014). We therefore replicated the same analysis
ical neurologist (A.P.L.) and defined by an automated lesion iden- after removing the lesion volume regressor.
tification algorithm (Seghier et al., 2008); (ii) 41 cm3 of right The presence of a lesion and lesion volume were based on
hemisphere damage and 51 cm3 of left hemisphere damage or binary lesion images generated by thresholding the fuzzy
41 cm3 of left hemisphere damage and 51 cm3 of right images. The threshold used to convert the fuzzy to binary
Table 1 Incidence of impaired performance for all right and left hemisphere stroke patients in Experiment 1
The Comprehensive Aphasia Test (CAT) RH LH RH LH

Sections Subtests Without Without With With
VPI VPI VPI VPI
n = 93 n = 307 n = 16 n = 62
I: The Cognitive Screen
1. Line bisection 0 0 12 23
2. Match pic-to-pic (semantic) 2 (2%) 6 (2%) 6 30
3. Recognition memory 3 (3%) 10 (3%) 3 31
Semantic memory Scorea 0 0 7 44
4. Word fluency 1 (1%) 90 (29%) 2 45
5. Gesture object use 2 (2%) 42 (14%) 3 21
6. Arithmetic 0 3 (1%) 1 11

II: The Language Battery
Comprehension 7. Match aud word-to-pic 2 (2%) 54 (18%) 8 32
8. Match auditory sentence-to-picture 12 (13%) 140 (46%) 9 57
9. Match auditory paragraph 0 44 (14%) 2 22
10. Match written word-to-picture 8 (9%) 91 (30%) 8 47
11. Match written sentence-to-picture 2 (2%) 109 (36%) 10 47
Repetition 12. Repetition of heard words 5 (5%) 151 (49%) 6 39
13. Repetition of complex words 4 (4%) 120 (39%) 4 35
14. Repetition of pseudowords 7 (8%) 104 (34%) 6 39
15. Repetition of digit strings 2 (2%) 125 (41%) 0 41
16. Repetition of sentences 1 (1%) 137 (45%) 1 45
Spoken output 17. Naming objects 5 (5%) 146 (48%) 6 47
18. Spoken picture description 7 (8%) 167 (54%) 11 52
Reading aloud 19. Reading words 3 (3%) 153 (50%) 8 44
20. Reading complex words 2 (2%) 133 (43%) 6 39
21. Reading function words 0 37 (12%) 0 22
22. Reading pseudowords 1 (1%) 142 (46%) 5 45
Writing 23. Copying letters 3 (3%) 26 (8%) 4 26
24. Written picture naming 0 61 (20%) 4 33
25. Writing to dictation 4 (4%) 125 (41%) 6 44
26. Written picture description 6 (6%) 154 (50%) 6 46
a
Semantic memory score is a combined score from picture-to-picture semantic matching and recognition memory. The action naming task was not included in the table because of
the high variability in scores across neurologically-normal controls (i.e. low specificity). The incidence of impaired performance is showed in absolute numbers and percentages.
LH = left hemisphere stroke patients; RH = right hemisphere stroke patients; VPI = visual perceptual impairments.
images was 0.3 as recommended in Seghier et al. (2008). Each Experiment 3: Are the regions iden-
binary lesion image was visually inspected by the operator.
The boundaries of the lesion may differ slightly from what is
tified in Experiment 2 involved in
seen by eye but provide an objective rather than subjective normal sentence comprehension?
measure of structural abnormality. There is no gold standard Twenty-five neurologically normal, native English speakers
of true abnormality. As a result classification errors are treated who were all right-handed according to the Edinburgh
as ‘noise’ in the analysis, biasing towards false negatives rather Handedness Inventory (Oldfield, 1971), were included in this
than false positives. functional MRI study, which aimed to identify whether the
The statistical output from the comparison of brain struc- right hemisphere regions from Experiment 2 were activated:
ture in the group of interest versus the control group was (i) during the task of interest from Experiment 1 (i.e. auditory
thresholded at P 5 0.05 after family-wise error (FWE) correc- sentence-to-picture matching; Supplementary Fig. 2); and
tion for multiple comparisons across the whole search volume (ii) by conditions that varied demands on auditory, visual,
(estimated using random field theory as implemented in SPM; phonological, semantic, sentence processing and verbal short-
Flandin and Friston, 2015). Having identified a significant term memory.
lesion-deficit mapping, we then examined the extent of the There were 10 conditions that each presented stimuli com-
effect at a voxel-level threshold of P 5 0.001 uncorrected, prising two pictures, auditory speech or both (Table 3) during
P 5 0.05 FWE-corrected cluster-level. All surviving voxels semantic matching, auditory repetition or speech retrieval
became our ‘VBM region’. Within this region, we report tasks. The 10 conditions contributed to three embedded fac-
the x, y, z MNI co-ordinates corresponding to the peak Z- torial designs (Supplementary Fig. 1). The first factorial design
scores. (Design A) combined six task conditions to compare sentences
Table 2 Demographic and clinical details for all right high definition pictures drawn by a professional artist). Each
hemisphere and left hemisphere stroke patients of the 120 objects was paired to three others making
3 60 = 180 pairs. The first pairing involved two objects inter-
Demographic and Full sample Without VPI acting with one another to indicate an event, with a corres-
clinical details ponding sentence (e.g. the cat is drinking from the jug). These
RH LH RH LH
n = 109 n = 369 n = 93 n = 307 were used for sentence production, sentence repetition,
verb naming and auditory sentence-to-picture matching. The
Age at scan, Mean 59.3 59.6 58.3 58.5
years SD 12.7 12.7 13.1 12.5 second pairing presented two unrelated objects (e.g. ‘car and
Minimum 23.1 21.3 23.1 21.3 plate’), that were used for object naming, auditory repetition,
Maximum 86.9 90.0 86.9 90.0 colour naming or auditory word to picture matching. The
Years since Mean 3.3 3.3 3.2 3.3 third pairing involved semantic pairs that were half related
stroke SD 2.3 2.5 2.1 2.4 (e.g. ‘door and key’) and half unrelated (e.g. ‘deer and
Minimum 0.3 0.3 0.3 0.3 barrel’). The three different pairings resulted in a total of
Maximum 9.2 10.0 9.2 10.0 180 different pairs (3 60). Stimulus repetitions, within sub-

Lesion size, Mean 63.4 76.0 49.0 67.8 ject, were avoided by repeating objects: (i) with a different
cm3 SD 81.7 82.7 63.8 75.2 pair; (ii) in different stimulus modalities (auditory versus
Minimum 1.1 1.0 1.1 1.0 visual or both); or (iii) with a change in task and response
Maximum 356.3 427.5 276.3 385.2 (matching versus spoken). Over participants each object was
Years of Mean 14.6 14.4 14.8 14.5 seen an equivalent number of times in each condition.
education SD 2.9 3.0 3.0 3.0 Compared to Experiment 1, auditory sentence-to-picture
Minimum 10.0 10.0 10.0 10.0 matching in Experiment 3 included: sentences with simpler
Maximum 22.0 26.0 22.0 26.0 structures (i.e. object-verb-subject) and only four possible
Gender, n Females 41 105 34 89 actions/verbs (jumping, falling, eating or drinking). The limited
Males 68 264 59 218 number of verbs was to minimize intersubject variability in
Self-report Difficulty 37 (93) 214 (291) 35 (83) 177 (244) word choice (or structure) during production.
assessment, n understanding
During sentence production, auditory repetition, object and
in first year
colour naming, participants were instructed to speak aloud in
Values in brackets indicate the number that had data on the self-assessment ques- the scanner so that we could distinguish correct and incorrect
tionnaire available. VPI = visual perceptual impairments. responses. Head movement was limited and corrected using
unwarping during image realignment. The degree of movement
made by each participant was monitored and it was not neces-
sary to remove any participants due to excessive movement-
to objects during: (i) speech-to-picture matching; (ii) auditory
related artefacts.
repetition; and/or (iii) speech retrieval. The second (Design B)
combined four naming conditions to isolate sentence process-
ing from the presence or absence of: (i) object names; and Functional MRI data acquisition and
(ii) verbs. The third (Design C) compared semantic associ-
ations to speech production tasks in the auditory and visual analysis
modalities (four conditions) while keeping stimuli constant Functional MRI data were acquired on a 3 T Trio scanner
within modality and the task constant across modality. (Siemens Medical Systems) using a 12-channel head coil and
In this third design, the interaction of task and stimulus mo- a gradient-echo EPI sequence with 3 3 mm in-plane reso-
dality tests the demands on: (i) phonological (name) retrieval, lution (repetition time/echo time/flip angle: 3080 ms/30 ms/90 ,
which is greater for speech production than semantic decisions extended field of view = 192 mm, matrix size = 64 64,
when the stimuli are pictures of objects than when the stimuli 44 slices, slice thickness = 2 mm, and interslice gap = 1 mm).
are heard object names (i.e. when the speech production task is Anatomical data were high resolution T1-weighted structural
picture naming rather than auditory repetition); and (ii) verbal images, acquired using exactly the same scanning parameters
short-term memory, which is greater for semantic decisions as for the patients imaged on the 3 T scanner (see above).
than speech production when the stimuli are auditory object Data preprocessing and statistical analyses were performed
names (which need to be held in memory while a semantic with the Statistical Parametric Mapping (SPM12) software
association is assessed) than when the stimuli are visually pre- package (Wellcome Centre for Human Neuroimaging,
sented objects (that do not require auditory memory). London UK; http://www.fil.ion.ucl.ac.uk/spm/). All functional
Each of the 10 tasks was presented in a separate scanning volumes were spatially realigned, unwarped, normalized to
session (counterbalanced across participants) with five blocks MNI space using the new unified normalization-segmentation
of four stimuli interleaved with 16.96 s of resting with eyes procedure, and smoothed with a 6 mm full-width half-
open (20 stimuli and 40 object concepts per condition). maximum isotropic Gaussian kernel, with a resulting voxel
Experimental and participant details are provided in Table 5. size of 3 3 3 mm.
Stimuli selection, creation and First-level analyses

counterbalancing Each preprocessed functional volume was entered into a sub-
Overall, we used a total of 120 object concepts that were easy ject-specific, fixed-effect analysis using the general linear model
to recognize and name when presented in picture format (using (Friston et al., 1995). All stimulus onset times, for all
2 Obj = pictures of two objects or two object names; Aud = auditory presentation of object names or sentences; Aud Rep = auditory repetition; Aud-Pic Match = matching an auditory stimulus to a picture; Inter = interaction between two
conditions, were modelled as single events (Mechelli et al.,
effects; PhR = phonological retrieval (highest for object naming/sentence production); Sem Assoc = matching two objects according to whether they are semantically related or not; Sent = sentences; SP = speech production; Vis = visual
2003). Stimuli with correct responses were modelled separately
Inter
Design C Stimulus/task vSTM/PhR
0
0
1
1
0
1
0
1
0
0
from stimuli with incorrect responses. Stimulus functions were
convolved with a canonical haemodynamic response function.
To exclude low-frequency confounds, the data were high-pass
Details of each of the 10 tasks (illustrated in Supplementary Fig. 1) and the weighting that each task was given in the factorial analysis of the three embedded designs (Designs A–C; see ‘Materials and methods’ section).
0
filtered using a set of discrete cosine basis functions with a cut-
0
1
1
0
1
0
1
0
0
SP
off period of 128 s. The contrasts of interest were generated for

each of the conditions (correct trials only) relative to fixation
baseline.
Sem
In Design A, Inter = effect of sentences 4 2 Obj on Aud-Pic Match 4 other tasks. In Design B, Inter = sentences (object names and verbs) 4 object names or verbs. In Design C, Inter = vSTM or PhR.
0
0
1
1
0
1
0
1
0
0
Second-level analyses
The contrasts from the first level analysis—one for each task
Aud
0
0
1
1
0
1
0
1
0
0
relative to rest—were entered into a 2 10 repeated measures

ANOVA. The second variable was intertrial interval, which
was 5 s for 12 subjects and 7 s for 13 subjects. This was
Design B Sentences Obj/Verbs
entered as a between-subjects factor with the 10 tasks as a

Inter
0
0
0
0
0
0
1
1
1
1
within-subjects factor. As there were no regions within the

search volume showing a main effect of intertrial interval (5 s
and 7 s intertrial interval groups) nor Intertrial interval Task
interactions, our results report the mean across both groups.
Verbs
The statistical threshold for second level contrasts was cor-

0
0
0
0
0
0
1
1
1
1
rected for multiple comparisons (i.e. family-wise error correc-

tion, FWE) within the lesion sites identified in Experiment 2
(Fig. 2A). Peak co-ordinates (x, y, z) are reported in MNI
2 Obj
space.
0
0
0
0
0
0
1
1
1
1
Experiment 4: Are the regions iden-

tified in Experiment 3 involved in
Design A Sentences ` 2 Object names
1
1
0
0
1
1
0
0
0
0
non-linguistic working memory?

Inter
Table 3 Experimental design and second level contrasts for Experiment 3
This functional MRI study investigated whether the right

hemisphere regions associated with the condition/task of inter-
est in Experiments 2 and 3, were activated when a new set of
presentation of pictures; vSTM = verbal short-term memory (highest for auditory semantic associations).
1
1
0
0
0
0
1
1
0
0
Second level contrasts
25 neurologically normal, right-handed, native English speak-

ers were making one-back matching decisions (i.e. is the stimu-
lus currently being displayed the same as the previously
presented stimulus?) on linguistic and non-linguistic stimuli.
Main effect
There were eight different conditions in a 2 2 2 factorial

design that manipulated: (i) auditory versus visual stimuli; with
(ii) speech versus non-speech content; and (iii) high versus low
1
1
0
0
1
1
1
1
0
0
semantic content (Table 4). In addition, the experimental para-

digm included eight corresponding speech production condi-
Finger/speech
tions that are not relevant to the current study but have
Response
been reported in Oberhuber et al. (2016), which investigated

SP colour
SP names
SP names
SP verbs
SP sent
SP sent
phonological processing in the left supramarginal gyrus.

Finger
Finger
Finger
Finger
Condition order was fully counterbalanced. Experimental

and participant details are provided in Table 5.
Pattern
The functional MRI acquisition, preprocessing and first-level

2 Obj
2 Obj
2 Obj
Event
Event
Event
See
analyses were exactly the same as described above for

Stimulus
-
-
-
Experiment 3. In the second-level analysis, we used a one-

2 Obj
2 Obj
2 Obj
Hear
way ANOVA, with eight contrasts, one for each one-back con-
Sent
Sent
dition relative to rest and report the main effect of semantic

-
-
-
-
-
content (words and objects versus pseudowords and meaning-

Produce Sentence
Name 2 Objects
Aud Sem Assoc
Aud Rep 2 Obj
less baselines), sublexical phonological cues (words/pseudo-

Name Colours
Vis Sem Assoc
Aud-Pic Match
Aud-Pic Match
Aud Rep Sent

Paradigm details
Produce Verb
Task name
words versus pictures and non-verbal sounds), the

interactions between these variables and with stimulus modality
(visual versus auditory). The search volume was restricted to
only include voxels that were part of two spheres (radius of
3 mm) centred on the peak co-ordinates obtained from the con-
ID
10
1
2
3
4
5
6
7
8
9
trast of interest in Experiment 3. We report effects that survived

Table 4 Experimental design and second level contrasts for Experiment 4
Paradigm details Second level contrasts

ID Task Stimulus Modality Semantic Sublexical phonology Interaction (semantic
contenta (semantic) (phonology) and phonology)
Visual Auditory Present Absent Present Absent Present Absent
O See pictures of objects 1 1 1 1 1 1 1 1
W See written object names 1 1 1 1 1 1 1 1
P See written pseudowords 1 1 1 1 1 1 1 1
B See coloured patterns 1 1 1 1 1 1 1 1
O Heard sounds of objects 1 1 1 1 1 1 1 1
W Heard names of objects 1 1 1 1 1 1 1 1
P Heard pseudowords 1 1 1 1 1 1 1 1

B Heard male/female voice 1 1 1 1 1 1 1 1
There were eight one-back matching with finger press response tasks that factorially manipulated the presence or absence of semantic content, the presence or absence of sublexical
phonology, using heard or written pseudowords (P), words (W) objects (O) or baselines (B). See text for details.
a
Task/condition of interest.
Table 5 Experimental details for Experiments 3 and 4

Results
Partcipants Experiment 3 Experiment 4
n 25 25 Experiment 1: Which language task is
Gender, n, females/males 15/10 12/13
Mean age in years ( SD) 30.4 (3.9) 31.4 (5.9)
most frequently impaired after right
Timing parameters hemisphere strokes?
Stimulus duration, s
Visual stimuli 2.5 1.5
The language task that was most frequently impaired after
Auditory stimuli/wordsa 1.8–2.5 0.64 right hemisphere stroke damage was auditory sentence-to-
Auditory pseudowords - 0.68 picture matching (Table 1 and Supplementary Table 3).
Intertrial interval, sb 5/7 2.5 Even after excluding all patients with visual perceptual def-
Block length, sc 20/28 22.5 icits, the incidence of impairments on the auditory sentence-
Total time for each run, mind 3.4/4.1 3.2 to-picture matching task was 13% (12/93) compared to
Total acquisition time, min 33.9/41.1 51.2 0–9% (mean = 4%) on all other language tasks (Table 1).
Number of stimuli per block 4 9 (incl. one
In contrast, in patients with left hemisphere stroke damage
repeat)
Number of blocks per run 5 4 the most frequently affected task (in those who did
Total number of stimuli per run 20 36 not have visual perceptual impairments) was spoken pic-
Number of runs 10 16 ture description, with an incidence of 54% (167/307)
Scanning parameters compared to: 1–50% (mean = 30%) on all other tasks
Repetition time, s 3.1 3.1 and 46% (140/307) on auditory sentence-to-picture match-
Number of slices 44 44 ing. The auditory sentence-to-picture matching and spoken
Number of volumes per run 61/85 62
picture description tasks are therefore the most sensitive
Number of dummy acquisitions 5 5
language comprehension and production indices, respect-
a
For Experiment 3, auditory stimuli included single words and sentences whereas for ively, in our language assessment.
Experiment 4 auditory stimuli included single words only.
b
To characterize differences in performance between lan-
For Experiment 3, 5 s intertrial interval = 5 s intertrial interval group (n = 12), 7 s
intertrial interval = 7 s intertrial interval group (n = 13). guage comprehension (i.e. auditory sentence-to-picture match-
c
For both functioanl MRI experiments, each block began with instructions for 3.1 s. ing task) and production (i.e. spoken picture description task)
d
For Experiment 3, each run ended with a resting period of 16.96/18.2 s for 5 s/7 s intertrial in left-hemisphere versus right-hemisphere stroke patients fur-
interval, respectively. For Experiment 4, each run ended with a resting period of 16 s.
ther, a 2 2 mixed factorial ANOVA was conducted on task
scores with Task (Production versus Comprehension) as a
a voxel-level threshold of P 5 0.05, after FWE-correction for
multiple comparisons within the regions of interest (Fig. 2B). within-subjects factor and Hemisphere Damaged (Left versus
Peak co-ordinates (x, y, z) are reported in MNI space. Right) as a between-subjects factor. We found a main effect
of Hemisphere Damaged [F(1,398) = 87.70, P 5 0.001], indi-
Data availability cating that left-hemisphere stroke patients (mean = 58.9) per-
formed, on average, significantly worse than right-hemisphere
The data that support the findings of this study are available
stroke patients (mean = 66.9). The main effect of Task was
from the senior author ([email protected]) upon reasonable
request. not significant [F(1,398) = 2.48, P = 0.116] but there was a
significant Hemisphere Task interaction [F(1,398) = 11.26, singer hits the soldier’), 7/9 made errors on the sentences
P = 0.001]. Post hoc tests confirmed that patients with unilat- with the simplest structures (e.g. ‘The woman is drinking’)
eral right-hemisphere lesions had poorer language comprehen- and 6/9 reported that their auditory speech comprehension
sion (mean = 65.8) than production (mean = 68.0; P = 0.005), had been compromised by their stroke (see Table 2 for the
while a trend in the opposite direction was observed in results of our self-report questionnaire that was conducted
patients with unilateral left-hemisphere lesions (mean without the patient’s knowledge of the test results).
Comprehension = 59.3 versus Production = 58.6; P = 0.066). Below, we systematically investigate whether their speech
To investigate the processing level that was affected in comprehension impairments might be the consequence of a
the 12 patients with right hemisphere damage and impaired reduction in the overall processing capacity available
scores on the auditory sentence-to-picture matching task, for syntactic, interpretive, and task-related operations
we considered how these patients performed on other lan- (Caplan et al., 2007), particularly in the auditory modality
guage tasks. We found that 9/12 of these patients were able (Thompson and Jefferies, 2013). This involved identifying
to: (i) repeat heard pseudowords and digit strings (two which right hemisphere regions were damaged in the nine

classic tests of speech perception and phonological working patients of interest with impairments in auditory sentence-
memory); and (ii) match visual sentences to pictures (that to-picture matching and how these regions respond to audi-
placed the same level of demand on semantic, syntactic and tory and visual linguistic and non-linguistic processing in
picture processing as auditory sentence-to-picture matching) neurologically normal individuals (Experiments 3 and 4).
(Supplementary Tables 1 and 2). They therefore had im-
pairments in auditory sentence-to-picture matching that Experiment 2: Which regions are
could not be explained by difficulties with speech percep- damaged in the patients of interest
tion, semantics, phonological working memory, syntactic
processing, or the integration of the syntactic structure of from Experiment 1?
a sentence with semantic information. The voxel-based lesion-deficit analysis compared the lesion
Examination of the frequency and type of errors that sites in our nine patients of interest to those of 75 control
these nine patients made during auditory sentence-to-pic- patients who did not have auditory sentence-to-picture
ture matching (Table 6) indicated that they had impaired matching impairments. This yielded one significant cluster
sentence comprehension: all nine made at least two errors (with 782 contiguous voxels) centred on the dorsal aspect
(either incorrect responses or delays/self-corrections) on the of the right superior longitudinal fasciculus (peak Z-
10 trials presenting difficult, reversible sentences (e.g. ‘The score = 6.7 at + 22, + 8, + 40) and impinging on the right
Table 6 Type of errors made by the nine patients of interest in the auditory sentence-to-picture matching task
Type of sentences Patient ID Stimulus

location
1 2 3 4 5 6 7 8 9
Non-reversible sentences
1. The woman is drinking - - 1 - - 0 - - 1 RU
2. The man is walking 1 - - - 1 0 1 - LB
3. She is laughing - - - - - - - - LB
4. The man is eating the apple - - - - - - - - RB
5. The woman is painting the wall - - - 1 - - 1 - - RU
6. The dog is sitting on the table - - - - - - - - - LU
Reversible sentences
7. The apple is under the shoe - - - - - - - - - RU
8. The nurse shoots the butcher 0 1 - 0 1 - - - - RU
9. The singer hits the soldier 0 0 0 0 0 1 0 0 0 LB
10. The policeman is painted by the dancer - - - - - 0 1 - 0 LU
11. The butcher is chased by the nurse - - - - 0 1 - - - RB
12. The dancer paints the policeman - - - - - - - - - LU
13. The shoe under the pencil is blue 0 0 0 1 1 - 1 0 1 RB
14. The carpet the cat is on is red - - 1 - - - - - - LU
15. The red pencil is under the shoe - - - - - - - - - LB
16. The flower in the cup is blue - - - - - - - - - RB
Type of errors made by the nine patients with impairments on the auditory sentence-to-picture matching task. 1 denotes a score of 1 for an accurate but delayed response, repetition
of the target by the examiner and/or self-correction. 0 denotes a score of 0 for incorrect responses. All other trials had a score of 2. Importantly, all incorrect responses on reversible
sentences corresponded to instances where the subject-verb-object relationship was reversed. For example, in Sentence 13, patients chose alternative (C): The pencil under the shoe
is blue. The last column shows the location in which the target sentences were displayed: R/L = right/left; B/U = bottom/upper. Patients had to select a picture, from a set of four
(2 2 array) that best illustrated the sentence that they heard. Patients 1–9 refer to the following IDs in the PLORAS database: PS0316, PS0383, PS0448, PS0670, PS0870, PS1172,
PS1211, PS1550 and PS2627, respectively.
inferior frontal sulcus (Z-score = 3.3 at + 32, + 4, + 34) small region A was 495% damaged in 3/9 patients of
(Fig. 1A). When the analysis was replicated without includ- interest and in 14/75 control patients. Put the other way,
ing lesion volume as a covariate of no-interest, virtually the 17 patients had 495% damage to region A and 3/17
same lesion-deficit associations were identified. Henceforth, (18%) of these patients had auditory sentence-to-picture
we focus on the results of the VBM analysis that factored matching impairments. Small region B was 495%
out linear effects from lesion size. damaged in 3/9 patients of interest and 5/75 control pa-
Post hoc analyses found that the ‘VBM region’ (i.e. 782 tients. Put the other way, eight patients had 495% damage
voxels in size) was substantially damaged (470%) in 6/9 to region B and 3/8 (38%) of these patients had auditory
of the patients of interest (67%) but only 3/75 (4%) of the sentence-to-picture matching impairments. Small region C
control patients. Put the other way, nine patients had was rare and only damaged in the patient of interest defin-
ing it.
470% damage to the VBM region and 6/9 (67%) of
these patients had auditory sentence-to-picture matching
impairments. In the remaining three patients of interest, Experiment 3: Are the regions iden-

the VBM region was completely preserved (0% damage). tified in Experiment 2 involved in
These three patients all had relatively small lesions affecting
parts of the putamen, thalamus, caudate or right temporal normal sentence comprehension?
lobe; see ‘small lesions’ A, B and C in Fig. 1B. Examination Within the region of interest (encompassing the VBM region
of how these regions were damaged across the two samples and small regions A and B from Experiment 2), we found
(patients of interest and control patients) indicated that significant activation for auditory sentence-to-picture
Figure 1 Right hemisphere lesion sites associated with impaired auditory sentence-to-picture matching. (A) The top row shows
the region identified in our voxel-based lesion-deficit analysis (VBM region in Experiment 2). (B) The bottom rows show the ‘small lesions’ A, B and
C from three patients with impaired auditory sentence-to-picture matching and no damage to the VBM region. Numbers below indicate the
corresponding MNI coordinates.

Figure 2 Illustration of right hemisphere activation in Experiments 3 and 4. (A) Top row: coronal slices show peak activations for
auditory sentence-to-picture matching relative to rest in the right inferior frontal sulcus (in red) and the right medial dorsal thalamus (in blue) at
coordinates [x = + 33, y = + 3, z = + 3] and [x = + 12, y = 9, z = + 9], respectively. White regions show the full extent of activation, after FWE
correction for multiple comparisons across the whole brain. Box plots depict medians with interquartile ranges and whiskers represent the 5th
and 95th percentiles. The black squares indicate the mean value for each task. Aud-Pic = auditory-to-picture matching tasks; C = colour naming; S
and O = sentences and objects; Sem Ass = semantic association tasks; V and A = visual and auditory presentation; Vb = verb (action) naming.
Numbers (bottom row) = condition number (Table 3 and Supplementary Fig. 1). (B) Bottom row: coronal slices showing peak activations for one-
back matching on pseudowords (P) and baselines (B) more than words (W) or pictures of objects (O) in the right inferior frontal sulcus (five
voxels in red) and the right medial dorsal thalamus (five voxels in blue) at coordinates [x = + 36, y = 0, z = + 33], Z score = 3.9, PFWE-corr = 0.001,
and [x = + 15, y = 9, z = + 6], Z score = 3.8, PFWE-corr = 0.001, respectively. White regions show the full extent of activation from whole brain
analysis; P 5 0.001, uncorrected. Box plots depict medians with interquartile ranges and whiskers represent the 5th and 95th percentiles. The
black squares indicate the mean value for each task.
matching (relative to rest) in the right inferior frontal sulcus other conditions (Fig. 2A) with no significant differences
(peak Z-score = 6.3 at + 33, + 3, + 33), and the right med- (P 4 0.001 uncorrected) in these regions for any of the ef-
iodorsal thalamus (peak Z-score = 4.3 at + 12, 9, + 9) fects of interest (Table 3) including: sentences compared to
(Fig. 2A). The same regions were also activated during all objects during any of the three task manipulations (Design
A); sentences, verbs or object names (Design B); or auditory participants performed language tasks and that activation
stimuli, semantic associations, verbal short-term memory, in these regions was more responsive to non-linguistic than
speech production or phonological (name) retrieval linguistic working memory demands. Below, we consider
(Design C). Nor were there any significant effects for the the results of our behavioural, lesion and functional ima-
reverse contrasts. ging analyses in the context of prior literature in order to
In summary, the results of Experiment 3 provide evidence demonstrate the scientific novelty, clinical implications and
that parts of the right hemisphere regions that were limitations of our findings.
damaged in patients with auditory sentence-to-picture
matching impairments (Experiment 2) are activated when
neurologically normal participants are matching auditory
Behavioural data: the language task
sentences to pictures but there was no evidence that these and processing level most frequently
regions were performing exclusively linguistic functions. affected by right hemisphere damage

Previous studies have reported that the incidence of
Experiment 4: Are the regions iden- acquired language disorders is 1–13% for right-handed
tified in Experiment 3 involved in right-hemisphere stroke patients (Alexander and Annett,
1996; Coppens et al., 2002) and 18–38% for right-
non-linguistic working memory? handed left-hemisphere stroke patients (Pedersen et al.,
We found that the right inferior frontal and thalamic re- 1995). The results from Experiment 1 are consistent with
gions, that were engaged by auditory sentence-to-picture these prior studies but also show how the incidence of lan-
matching in Experiment 3 and damaged in patients with guage impairments is task-dependent, even after controlling
auditory sentence-to-picture matching in Experiment 2, for visual perceptual abilities. For patients with unilateral
were significantly more activated (Z scores = 3.9 and 3.8, right hemisphere damage, the highest incidence of impaired
respectively) when one-back matching was performed in performance (13%) was recorded for auditory sentence-to-
the absence of semantic information (i.e. for pseudowords picture matching, which tests spoken sentence comprehen-
and baselines compared to word and object stimuli) sion abilities. This cannot simply be explained in terms of
(Fig. 2B). There were no significant effects in any part of task difficulty, because, in patients with left hemisphere
the right hemisphere that could be attributed to the de- damage, the most frequently impaired task was spoken pic-
mands on phonological or semantic processing or the inter- ture description. Our findings are, therefore, consistent with
action between semantic and phonological processing. prior literature in post-stroke aphasia showing that the
In other words, the response in the right hemisphere, right hemisphere might contribute to speech comprehension
including our regions of interest, was more consistent more than speech production (Zaidel, 1976; Crinion and
with non-linguistic than linguistic processing demands. Price, 2005).
By examining how the patients with right hemisphere
damage and impaired auditory sentence-to-picture match-
ing performed on other cognitive and language tasks, we
Discussion identified a group of nine patients who were not impaired
By using a methodological approach that integrates behav- on tasks that collectively place similar demands on visual
ioural, lesion and functional imaging data, we argue below and auditory perception, phonological, semantic and syn-
that speech comprehension can be impaired after right tactic processing and verbal short-term memory. We there-
hemisphere stroke because: (i) normal speech comprehen- fore hypothesized that their difficulty matching auditory
sion increases the demands on non-linguistic working sentences to pictures might be related to instances when
memory; and (ii) non-linguistic working memory (an execu- word order needed to be held in memory or the task
tive function) is supported by right hemisphere regions. placed high demands on executive (working memory) func-
In brief, the behavioural data allowed us to identify a tions. These types of processing may be more demanding
group of patients who had right hemisphere damage and during auditory than visual sentence-to-picture matching,
poor scores on one or more language tasks, and generate because the auditory sentences are only heard once,
hypotheses to explain what level of processing impairment before decisions and responses are required, whereas the
(e.g. perceptual, semantic, syntactic or executive) might patient can continue reading the sentence while making a
underlie their poor language scores. The lesion analyses decision with written sentence-to-picture matching.
enabled us to create regions of interest by comparing the Although it was not possible to assess the patient’s deficits
lesion sites in patients with right hemisphere damage and further, we conclude that their difficulties with auditory
poor language task scores to the lesion sites in other pa- sentence-to-picture matching were more likely to be the
tients who had right hemisphere damage in the absence of consequence of disrupted executive processing than impair-
impaired language task scores. Finally, functional imaging ments in linguistic or perceptual processing. It is also pos-
allowed us to show that parts of the regions identified in sible that mild executive-semantic impairments, paired with
the lesion study were activated when neurologically normal disrupted connectivity from auditory input, gives rise to
semantic ‘access’ deficits affecting the auditory modality right hemisphere responds to language and executive pro-
only (Thompson and Jefferies, 2013). cessing in neurologically normal participants.
Abnormally low auditory sentence-to-picture matching
scores, in the context of good perceptual skills (as observed
Functional imaging data: the contri-
in our nine patients of interest) are likely to reflect impaired
speech comprehension in everyday conversations, even if bution of the identified right hemi-
the patients were not fully aware of their own limitations. sphere regions to normal speech
This is because the auditory sentence-to-picture matching comprehension
task includes simple constructions (‘The woman is drink-
ing’) plausibly encountered in everyday speech as well as Formal evidence that the right hemisphere is normally
ones (‘The flower in the cup is blue’) mirrored in everyday involved in matching spoken sentences-to-pictures is pro-
relative clauses (‘And the plans that are available to us vided by Experiment 3. Within the regions that were
range from kind of mediocre to really sweet’; cited in damaged in patients with auditory sentence-to-picture

Roland et al., 2007). matching impairments, activation was observed in the
right inferior frontal sulcus and right mediodorsal thal-
amus. In addition, these regions responded during a range
Lesion analyses: the right hemisphere of language tasks, with no evidence to suggest that they
were particularly responsive to perceptual, semantic,
lesion sites associated with impaired phonological or syntactic processing. The results of a
auditory sentence-to-picture second functional imaging experiment (Experiment 4) ex-
matching plain this finding by showing that the right inferior frontal
sulcus and right mediodorsal thalamus are sensitive to the
In Experiment 2, we found that the right hemisphere re-
demands on non-linguistic (i.e. domain-general) working
gions that were most frequently damaged in patients with
memory capacity because they were significantly more acti-
impaired auditory sentence-to-picture matching included
vated when one-back matching was performed on stimuli
dorsal parts of the superior longitudinal fasciculus imping- that lacked semantic content (Fig. 2B). Our data thus com-
ing on the right inferior frontal sulcus, and more ventral plement and extend the results of previous studies of se-
subcortical regions in the vicinity of the right putamen, mantic cognition (Jefferies, 2013; Thompson et al., 2016)
thalamus and caudate (Fig. 1). Damage to these right hemi- that have shown that regions in the right middle cerebral
sphere regions was: (i) frequently observed (32% had sub- artery territory contribute to executive aspects of semantic
stantial damage to at least one of these regions); and processing (i.e. controlled semantic retrieval). Using func-
(ii) not infrequently associated with impaired auditory sen- tional imaging of neurologically normal participants, we
tence-to-picture matching (e.g. 67% of those with damage show the most critical region is likely to be the right infer-
to the VBM region had impaired auditory sentence-to- ior frontal sulcus and that the function of this region is not
picture matching when tested months after their stroke). limited to semantic tasks.
Inter-patient variability in the effects of lesions to regions Our neuropsychological, lesion and functional MRI data
that show highly significant effects in group-level voxel- therefore collectively support the hypothesis that difficulties
based analyses has also been observed in studies of patients performing the auditory sentence-to-picture matching task
with left hemisphere damage. For example, in Gajardo- after right hemisphere damage could result from disruption
Vidal et al. (2018), we found that the incidence of long- to non-linguistic executive processing that is necessary for
term lexical retrieval impairments following damage to normal language function.
regions identified in group-level voxel-based analyses
(with very conservative statistical thresholds) was 550%.
If inter-patient variability is due to differences in the ability Scientific novelty
to recover, future studies should find that variability is less Our findings link three unrelated observations in the prior
when patients are tested in the acute stage of stroke before literature: (i) right inferior frontal and right mediodorsal
recovery from initial deficits occurs. The lesion-deficit asso- thalamus activity increase during executive processing;
ciation might also be more consistent across subjects if the (ii) executive processing is required for sentence compre-
tests used more sensitive measures of impairments such as hension; and (iii) right frontal activity increases during sen-
reaction times (which are not currently available from our tence comprehension.
assessments). The role of the right inferior frontal and right mediodor-
Although further studies are required to understand sal thalamus in executive functions (e.g. planning, monitor-
which patients are more versus less affected by right hemi- ing, switching and inhibition) has been demonstrated in
sphere damage, the key point here is that the effect of many prior studies (Aron et al., 2004, 2014; Halassa and
damage to these right hemisphere sites was not atypical. Kastner, 2017; Neef et al., 2018). For example, neuropsy-
The results of the lesion analyses were therefore used to chological studies have reported difficulties in working
provide regions of interest for an investigation of how the memory and inhibitory control after focal damage to
both the right frontal lobe (Szczepanski and Knight, 2014) disruption to normal functional anatomy. Third, we experi-
and the right mediodorsal thalamus (Van der Werf et al., mentally link the literature on three unrelated topics (sum-
2003); and multiple functional MRI studies have reported marized above) by identifying a right inferior frontal region
activation changes in right inferior frontal cortex (Sebastian that is (i) damaged in patients who have auditory sentence-
et al., 2016) and right mediodorsal thalamus (Andrews to-picture matching impairments; (ii) activated when neuro-
et al., 2006; Minzenberg et al., 2009) under a variety of logically normal participants are performing auditory sen-
test conditions that tax executive functions. Evidence that tence-to-picture matching; and (iii) sensitive to the demands
these two regions work as part of a single executive system on non-linguistic and linguistic working memory. This pro-
is further provided by anatomical connectivity studies that vides the first evidence that the same right hemisphere re-
have identified reciprocal fronto-thalamic connections gions are contributing to both sentence comprehension and
(Behrens et al., 2003; Hwang et al., 2010; Eckert et al., executive function.
2012; Jeon et al., 2014).
The importance of good executive functions for speech Clinical relevance

comprehension has been shown by manipulating ambiguity
in the sentence content (Key-DeLyria and Altmann, 2016). Our study shows that damage to the right inferior frontal
For example, individuals with higher IQ scores and faster cortex and right mediodorsal thalamus can impair sentence
processing were more likely to answer ambiguous sentence comprehension, and this is likely to be the consequence of
comprehension questions correctly (Engelhardt et al., disruption to a right hemisphere system that is involved in
2017), and older adults with good inhibition skills normal language processing. Our conclusion complements
showed better sentence comprehension than those with findings from patients with semantic dementia, which high-
poor inhibition skills (Yoon et al., 2015). Interestingly, syn- light the role of the right anterior temporal lobe in semantic
tactic interference effects during sentence comprehension representations (Rice et al., 2015; Jung and Lambon Ralph,
were found to be predicted by general working memory 2016; Thompson et al., 2016; Lambon Ralph et al., 2017).
capacity but not by phonological memory capacity as mea- However, we are not claiming that language impairments
sured by digit span (Tan et al., 2017). This highlights a role following right hemisphere strokes can always be explained
for non-linguistic working memory in sentence comprehen- by a disruption to parts of the normal language system.
sion that is over and above the contribution of verbal Other patients may present with impaired language abilities
working memory capacity and may explain why our pa- after right hemisphere damage as a result of atypical
tients were not found to have abnormally low digit spans. language lateralization (i.e. crossed aphasia; Marien et al.,
The role of the right inferior frontal lobe in sentence 2004).
comprehension has also been shown previously, particu-
larly for older compared to younger neurologically Limitations, clarifications and future
normal participants (Wingfield and Grossman, 2006),
when words are ambiguous (Mason and Just, 2007),
directions
when sentences are reversible (Meltzer et al., 2010) or in- Our study has focused on a carefully selected group of
determinate (de Almeida et al., 2016), and when patients patients who had the most consistent language impairment
with aphasia after left hemisphere strokes are recovering after right hemisphere stroke and demonstrated that this
their sentence comprehension abilities (van Oers et al., language impairment can be explained by disruption to
2010; Mohr et al., 2014; Kielar et al., 2016). Finally, im- normal functional anatomy. However, we are not exclud-
plication that the right hemisphere may be playing a non- ing the possibility that language difficulties in other patients
linguistic executive role in normal speech comprehension with right hemisphere damage can be a consequence of
has also been proposed (Bozic et al., 2010; Vigneau atypical hemispheric lateralization of language prior to
et al., 2011; Baumgaertner et al., 2013). their stroke. Future studies are needed to demonstrate
Together, these studies provide abundant evidence that this formally by combining neuropsychological data
non-linguistic executive processing in the right hemisphere with functional MRI investigations of neurologically typical
is important for speech comprehension. Nonetheless, we participants.
are not dismissing the potential contribution of right hemi- In neurologically normal participants, we identified a right
sphere regions to language processing itself. For instance, inferior frontal region and a right mediodorsal thalamic
there is accumulating evidence showing that bilateral anter- region that were both activated during language tasks
ior temporal lobes are involved in the representation of (Experiment 3) and also when the demands on non-linguistic
conceptual knowledge (Rice et al., 2015; Jung and working memory increased (Experiment 4). Future neuro-
Lambon Ralph, 2016; Lambon Ralph et al., 2017). psychological studies are needed to assess whether damage
Our study adds to previous literature in several ways. to these regions impairs non-linguistic (domain-general) ex-
First, we demonstrate that damage to the right inferior ecutive functions as well as sentence comprehension.
frontal sulcus and right mediodorsal thalamus can impair Three of nine patients of interest were not aware that
spoken sentence comprehension. Second, we show that the their everyday speech comprehension had been compro-
effect of damage to these regions can be explained by mised by their stroke. Their lack of awareness is not
unexpected given that, in a previous study, we found that

patients with comprehension difficulties are less aware of Funding
their impairments than patients with speech production dif- This work was supported by Wellcome (203147/Z/16/Z
ficulties (unpublished results). However, future studies are and 205103/Z/16/Z, C.J.P.), the Medical Research
needed to investigate sentence comprehension and self- Council (MR/M023672/1, C.J.P and M.L.S.) and the
awareness of comprehension abilities in more detail and Stroke Association (TSA 2014/02, C.J.P. and D.W.G.;
to establish how these measurements change over time TSA PDF 2017/02, T.M.H.H.). A.G.-V. (CONICYT
post-stroke. BECAS-CHILE 73101009) and D.L.L.-P. (CONICYT
Finally, we note that, although the functional MRI experi- BECAS-CHILE 72140131) were funded by the Chilean
ments focus on the function of the grey matter regions, we National Commission for Scientific and Technological
are not dismissing the likely contribution of the surrounding Research (CONICYT) through its scholarship scheme for
white matter, which is expected to play the important role of graduate studies abroad.
propagating activity to and from other task-related regions.

Competing interests
Conclusion The authors report no competing interests.
By combining behavioural, lesion and functional MRI data,
our results demonstrate that: (i) long-lasting speech com-
prehension impairments were frequently observed in our
right-handed patients with right inferior frontal damage;
Supplementary material
(ii) this can be explained by disruption to normal functional Supplementary material is available at Brain online.
anatomy rather than being indicative of crossed aphasia/
atypical language lateralization; and (iii) the same right
hemisphere regions contribute to both sentence comprehen-
sion and executive functions. Seemingly domain-specific References
cognitive deficits (e.g. language processing) can therefore Alexander MP, Annett M. Crossed aphasia and related anomalies of
be explained by disruptions to domain-general cognitive cerebral organization: case reports and a genetic hypothesis. Brain
mechanisms (e.g. non-linguistic executive processing). Lang 1996; 55: 213–39.
Andrews J, Wang L, Csernansky JG, Gado MH, Barch DM.
Conversely, these findings are consistent with previous re-
Abnormalities of thalamic activation and cognition in schizophrenia.
ports that attribute increased right inferior frontal activity Am J Psychiatry 2006; 163: 463–9.
during speech comprehension, in patients with left hemi- Aron AR, Robbins TW, Poldrack RA. Inhibition and the right inferior
sphere lesions, to a greater reliance on domain-general cog- frontal cortex. Trends Cogn Sci 2004; 8: 170–7.
nitive processing (van Oers et al., 2010; Geranmayeh et al., Aron AR, Robbins TW, Poldrack RA. Inhibition and the right inferior
frontal cortex: one decade on. Trends Cogn Sci 2014; 18: 177–85.
2014).
Ashburner J, Friston KJ. Voxel-based morphometry-the methods.
Clinically, our study highlights the possibility that speech Neuroimage 2000; 11: 805–21.
comprehension difficulties after right hemisphere stroke Bartolomeo P, Thiebaut de Schotten M, Chica AB. Brain networks of
may be long-lasting and not due to atypical language lat- visuospatial attention and their disruption in visual neglect. Front
eralization. Given that around 10 million people worldwide Hum Neurosci 2012; 6: 110.
Baumgaertner A, Hartwigsen G, Roman Siebner H. Right-hemispheric
survive a stroke each year (Feigin et al., 2017), the number
processing of non-linguistic word features: implications for
of patients with speech comprehension difficulties after mapping language recovery after stroke. Hum Brain Mapp 2013;
right hemisphere damage is likely to be substantial, in 34: 1293–305.
right-handed as well as left-handed patients. Clinical Behrens TEJ, Johansen-Berg H, Woolrich MW, Smith SM, Wheeler-
advice should therefore alert patients with right hemisphere Kingshott CAM, Boulby PA, et al. Non-invasive mapping of con-
nections between human thalamus and cortex using diffusion ima-
stroke to the possibility that they may suffer from long- ging. Nat Neurosci 2003; 6: 750–7.
lasting speech comprehension difficulties, rather than in- Bozic M, Tyler LK, Ives DT, Randall B, Marslen-Wilson WD.
forming them that right hemisphere strokes do not typically Bihemispheric foundations for human speech comprehension. Proc
cause long-term language difficulties. Natl Acad Sci USA 2010; 107: 17439–44.
Butler RA, Lambon Ralph MA, Woollams AM. Capturing multidi-
mensionality in stroke aphasia: mapping principal behavioural com-
ponents to neural structures. Brain 2014; 137: 3248–66.
Acknowledgements Caplan D, Waters G, Dede G, Michaud J, Reddy A. A study of syn-
tactic processing in aphasia I: behavioral (psycholinguistic) aspects.
We would like to thank the PLORAS recruitment team Brain Lang 2007; 101: 103–50.
Coppens P, Hungerford S, Yamaguchi S, Yamadori A. Crossed apha-
(http://www.ucl.ac.uk/ploras) for their help with data col- sia: an analysis of the symptoms, their frequency, and a comparison
lection. We are indebted to the patients and their carers for with left-hemisphere aphasia symptomatology. Brain Lang 2002; 83:
their generous assistance with our research. 425–63.
Corbetta M, Kincade MJ, Lewis C, Snyder AZ, Sapir A. Neural basis Jung J, Lambon Ralph MA. Mapping the dynamic network inter-
and recovery of spatial attention deficits in spatial neglect. Nat actions underpinning cognition: a cTBS-fMRI study of the flexible
Neurosci 2005; 8: 1603–10. adaptive neural system for semantics. Cereb Cortex 2016; 26: 3580–
Crinion J, Price CJ. Right anterior superior temporal activation pre- 90.
dicts auditory sentence comprehension following aphasic stroke. Key-DeLyria SE, Altmann LJP. Executive function and ambiguous
Brain 2005; 128: 2858–71. sentence comprehension. Am J Speech Lang Pathol 2016; 25:
de Almeida RG, Riven L, Manouilidou C, Lungu O, Dwivedi VD, 252–67.
Jarema G, et al. The neuronal correlates of indeterminate sen- Kielar A, Deschamps T, Jokel R, Meltzer JA. Functional reorganiza-
tence comprehension: an fMRI Study. Front Hum Neurosci 2016; tion of language networks for semantics and syntax in chronic
10: 614. stroke: evidence from MEG. Hum Brain Mapp 2016; 37: 2869–93.
Eckert U, Metzger CD, Buchmann JE, Kaufmann J, Osoba A, Li M, Lambon Ralph M, Jefferies E, Patterson K, Rogers TT. The neural and
et al. Preferential networks of the mediodorsal nucleus and centro- computational bases of semantic cognition. Nat Rev Neurosci 2017;
median-parafascicular complex of the thalamus-a DTI tractography 18: 42–55.
study. Hum Brain Mapp 2012; 33: 2627–37. Lorca-Puls DL, Gajardo-Vidal A, White J, Seghier ML, Leff AP, Green
Engelhardt PE, Nigg JT, Ferreira F. Executive function and intelligence DW, et al. The impact of sample size on the reproducibility of voxel-
in the resolution of temporary syntactic ambiguity: an individual dif- based lesion-deficit mappings. Neuropsychologia 2018; 115: 101–

ferences investigation. Q J Exp Psychol (Hove) 2017; 70: 1263–81. 11.
Feigin VL, Norrving B, Mensah GA. Global burden of stroke. Circ Res Marien P, Paghera B, Dedeyn P, Vignolo L. Adult crossed aphasia in
2017; 120: 439–48. dextrals revisited. Cortex 2004; 40: 41–74.
Flandin G, Friston KJ. Topological inference A2 - Toga, Arthur W. Mason RA, Just MA. Lexical ambiguity in sentence comprehension.
Brain mapping. Waltham: Academic Press; 2015. p. 495–500. Brain Res 2007; 1146: 115–27.
Forkel SJ, Thiebaut de Schotten M, Dell’Acqua F, Kalra L, Murphy Mechelli A, Henson RNA, Price CJ, Friston KJ. Comparing event-
DG, Williams SC, et al. Anatomical predictors of aphasia recovery: related and epoch analysis in blocked design fMRI. Neuroimage
a tractography study of bilateral perisylvian language networks. 2003; 18: 806–10.
Brain 2014; 137: 2027–39. Meltzer JA, McArdle JJ, Schafer RJ, Braun AR. Neural aspects of
Fridriksson J, Yourganov G, Bonilha L, Basilakos A, Den Ouden D-B, sentence comprehension: syntactic complexity, reversibility, and re-
Rorden C. Revealing the dual streams of speech processing. Proc analysis. Cereb Cortex 2010; 20: 1853–64.
Natl Acad Sci USA 2016; 113: 15108–13. Minzenberg MJ, Laird AR, Thelen S, Carter CS, Glahn DC. Meta-
Friston KJ, Frith CD, Turner R, Frackowiak RS. Characterizing analysis of 41 functional neuroimaging studies of executive function
evoked hemodynamics with fMRI. Neuroimage 1995; 2: 157–65. in schizophrenia. Arch Gen Psychiatry 2009; 66: 811–22.
Gajardo-Vidal A, Lorca-Puls DL, Crinion J, White J, Seghier ML, Leff Mohr B, Difrancesco S, Harrington K, Evans S, Pulvermuller F.
AP, et al. How distributed processing produces false negatives in Changes of right-hemispheric activation after constraint-induced, in-
voxel-based lesion-deficit analyses. Neuropsychologia, 2018; 115: tensive language action therapy in chronic aphasia: fMRI evidence
124–33 from auditory semantic processing. Front Human Neurosci 2014; 8:
Geranmayeh F, Brownsett SLE, Wise RJS. Task-induced brain activity 919.
in aphasic stroke patients: what is driving recovery? Brain 2014; Mummery CJ, Patterson K, Price CJ, Ashburner J, Frackowiak RS,
137: 2632–48. Hodges JR. A voxel-based morphometry study of semantic demen-
Halassa MM, Kastner S. Thalamic functions in distributed cognitive tia: relationship between temporal lobe atrophy and semantic
control. Nat Neurosci 2017; 20: 1669–79. memory. Ann Neurol 2000; 47: 36–45.
Hartwigsen G, Baumgaertner A, Price CJ, Koehnke M, Ulmer S, Nardo D, Holland R, Leff AP, Price CJ, Crinion JT. Less is more:
Siebner HR. Phonological decisions require both the left and right neural mechanisms underlying anomia treatment in chronic aphasic
supramarginal gyri. Proc Natl Acad Sci USA 2010a; 107: 16494–9. patients. Brain 2017; 14: 3039–54.
Hartwigsen G, Price CJ, Baumgaertner A, Geiss G, Koehnke M, Ulmer Neef NE, Anwander A, Bütfering C, Schmidt-Samoa C, Friederici AD,
S, et al. The right posterior inferior frontal gyrus contributes to Paulus W, et al. Structural connectivity of right frontal hyperactive
phonological word decisions in the healthy brain: evidence from areas scales with stuttering severity. Brain, 2018; 141: 191–204.
dual-site TMS. Neuropsychologia 2010b; 48: 3155–63. Oberhuber M, Hope TM, Seghier ML, Parker Jones O, Prejawa S,
Hartwigsen G, Saur D, Price CJ, Ulmer S, Baumgaertner A, Siebner Green DW, et al. Four functionally distinct regions in the left supra-
HR. Perturbation of the left inferior frontal gyrus triggers adaptive marginal gyrus support word processing. Cereb Cortex 2016; 26:
plasticity in the right homologous area during speech production. 4212–26.
Proc Natl Acad Sci USA 2013; 110: 16402–7. Oldfield RC. The assessment and analysis of handedness: the
Hillis AE, Newhart M, Heidler J, Barker PB, Herskovits EH, Edinburgh inventory. Neuropsychologia 1971; 9: 97–113.
Degaonkar M. Anatomy of spatial attention: insights from perfusion Pedersen PM, Jorgensen HS, Nakayama H, Raaschou HO, Olsen TS.
imaging and hemispatial neglect in acute stroke. J Neurosci 2005; Aphasia in acute stroke: incidence, determinants, and recovery. Ann
25: 3161–7. Neurol 1995; 38: 659–66.
Hwang K, Velanova K, Luna B. Strengthening of top-down frontal Price CJ, Crinion JT, Leff AP, Richardson FM, Schofield TM, Prejawa
cognitive control networks underlying the development of inhibitory S, et al. Lesion sites that predict the ability to gesture how an object
control: a functional magnetic resonance imaging effective connect- is used. Arch Ital Biol 2010; 148: 243–58.
ivity study. J Neurosci 2010; 30: 15535–45. Ravizza SM, Behrmann M, Fiez JA. Right parietal contributions to
Jefferies E. The neural basis of semantic cognition: converging evidence verbal working memory: spatial or executive? Neuropsychologia
from neuropsychology, neuroimaging and TMS. Cortex 2013; 49: 2005; 43: 2057–67.
611–25 Rice GE, Lambon Ralph MA, Hoffman P. The roles of left versus right
Jeon HA, Anwander A, Friederici AD. Functional network mirrored in anterior temporal lobes in conceptual knowledge: an ALE meta-ana-
the prefrontal cortex, caudate nucleus, and thalamus: high-reso- lysis of 97 functional neuroimaging studies. Cereb Cortex 2015; 25:
lution functional imaging and structural connectivity. J Neurosci 4374–91.
2014; 34: 9202–12. Roland D, Dick F, Elman JL. Frequency of basic english grammatical
Jonides J, Smith EE, Koeppe RA, Awh E, Minoshima S, Mintun MA. structures: a corpus analysis. J Mem Lang 2007; 57: 348–79
Spatial working memory in humans as revealed by PET. Nature Sebastian A, Jung P, Neuhoff J, Wibral M, Fox PT, Lieb K, et al.
1993; 363: 623–5. Dissociable attentional and inhibitory networks of dorsal and
ventral areas of the right inferior frontal cortex: a combined task- Thompson HE, Henshall L, Jefferies E. The role of the right hemi-
specific and coordinate-based meta-analytic fMRI study. Brain Struct sphere in semantic control: a case-series comparison of right and left
Funct 2016; 221: 1635–51. hemisphere stroke. Neuropsychologia 2016; 85: 44–61.
Seghier ML, Patel E, Prejawa S, Ramsden S, Selmer A, Lim L, et al. Thompson HE, Jefferies E. Semantic control and modality: an input
The PLORAS database: a data repository for predicting language processing deficit in aphasia leading to deregulated semantic cogni-
outcome and recovery after stroke. Neuroimage 2016; 124: 1208– tion in a single modality. Neuropsychologia 2013; 51: 1998–2015.
12. Van der Werf YD, Scheltens P, Lindeboom J, Witter MP, Uylings HB,
Seghier ML, Ramlackhansingh A, Crinion J, Leff AP, Price CJ. Lesion Jolles J. Deficits of memory, executive functioning and attention
identification using unified segmentation-normalisation models and following infarction in the thalamus; a study of 22 cases with loca-
fuzzy clustering. Neuroimage 2008; 41: 1253–66. lised lesions. Neuropsychologia 2003; 41: 1330–44.
Sollmann N, Tanigawa N, Ringel F, Zimmer C, Meyer B, Krieg SM. van Oers CA, Vink M, van Zandvoort MJ, van der Worp HB, de
Language and its right-hemispheric distribution in healthy brains: an Haan EH, Kappelle LJ, et al. Contribution of the left and right
investigation by repetitive transcranial magnetic stimulation. inferior frontal gyrus in recovery from aphasia. A functional MRI
Neuroimage 2014; 102: 776–88. study in stroke patients with preserved hemodynamic responsive-
Sperber C, Karnath HO. Impact of correction factors in human ness. Neuroimage 2010; 49: 885–93.
brain lesion-behavior inference. Hum Brain Mapp 2017; 38: Vigneau M, Beaucousin V, Herve PY, Jobard G, Petit L, Crivello F,

1692–701. et al. What is right-hemisphere contribution to phonological, lexico-
Swinburn K, Porter G, Howard D. Comprehensive aphasia test. Hove: semantic, and sentence processing? Insights from a meta-analysis.
Psychology Press; 2004. Neuroimage 2011; 54: 577–93.
Szczepanski SM, Knight RT. Insights into human behavior from le- Wingfield A, Grossman M. Language and the aging brain: patterns of
sions to the prefrontal cortex. Neuron 2014; 83: 1002–18. neural compensation revealed by functional brain imaging. J
Tan Y, Martin RC, Van Dyke JA. Semantic and syntactic interference Neurophysiol 2006; 96: 2830–9.
in sentence comprehension: a comparison of working memory Xing S, Lacey EH, Skipper-Kallal LM, Jiang X, Harris-Love ML, Zeng
models. Front Psychol 2017; 8: 198. J, et al. Right hemisphere grey matter structure and language out-
Thiel A, Habedank B, Herholz K, Kessler J, Winhuisen L, Haupt WF, comes in chronic left hemisphere stroke. Brain 2016; 139: 227–41.
et al. From the left to the right: how the brain compensates progres- Yoon J, Campanelli L, Goral M, Marton K, Eichorn N, Obler LK.
sive loss of language function. Brain Lang 2006; 98: 57–65. The effect of plausibility on sentence comprehension among older
Thompson HE, Almaghyuli A, Noonan KA, Barak O, Lambon Ralph adults and its relation to cognitive functions. Exp Aging Res 2015;
MA, Jefferies E. The contribution of executive control to semantic 41: 272–302.
cognition: convergent evidence from semantic aphasia and executive Zaidel E. Auditory vocabulary of the right hemisphere following brain
dysfunction. J Neuropsychol 2018; 12: 312–40 bisection or hemidecortication. Cortex 1976; 12: 191–211.

Hemisferio Derecho Comprension

Uploaded by

Copyright:

Available Formats

Hemisferio Derecho Comprension

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hemisferio Derecho Comprension

Uploaded by

Copyright:

Available Formats

doi:10.

1093/brain/awy270 BRAIN 2018: 141; 3389–3404 | 3389

How right hemisphere damage after stroke

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Correspondence to: Andrea Gajardo-Vidal

expect the regions that are damaged in the language im-

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

The Comprehensive Aphasia Test (CAT) RH LH RH LH

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Stimuli selection, creation and First-level analyses

off period of 128 s. The contrasts of interest were generated for

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

entered as a between-subjects factor with the 10 tasks as a

within-subjects factor. As there were no regions within the

The statistical threshold for second level contrasts was cor-

rected for multiple comparisons (i.e. family-wise error correc-

Experiment 4: Are the regions iden-

non-linguistic working memory?

This functional MRI study investigated whether the right

25 neurologically normal, right-handed, native English speak-

There were eight different conditions in a 2 2 2 factorial

semantic content (Table 4). In addition, the experimental para-

been reported in Oberhuber et al. (2016), which investigated

phonological processing in the left supramarginal gyrus.

Condition order was fully counterbalanced. Experimental

The functional MRI acquisition, preprocessing and ﬁrst-level

analyses were exactly the same as described above for

Experiment 3. In the second-level analysis, we used a one-

dition relative to rest and report the main effect of semantic

content (words and objects versus pseudowords and meaning-

Aud Rep 2 Obj

less baselines), sublexical phonological cues (words/pseudo-

Aud Rep Sent

words versus pictures and non-verbal sounds), the

trast of interest in Experiment 3. We report effects that survived

Table 4 Experimental design and second level contrasts for Experiment 4

Paradigm details Second level contrasts

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Table 5 Experimental details for Experiments 3 and 4

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Type of sentences Patient ID Stimulus

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

unexpected given that, in a previous study, we found that

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

Downloaded from https://academic.oup.com/brain/article/141/12/3389/5168496 by guest on 17 April 2023

You might also like