1 s2.0 S1053811916301719 Main

NeuroImage 137 (2016) 165–177
Contents lists available at ScienceDirect
NeuroImage
journal homepage: www.elsevier.com/locate/ynimg
Exploring the role of the posterior middle temporal gyrus in semantic

cognition: Integration of anterior temporal lobe with executive processes
James Davey a, Hannah E. Thompson a, Glyn Hallam a, Theodoros Karapanagiotidis a, Charlotte Murphy a,
Irene De Caso a, Katya Krieger-Redwood a, Boris C. Bernhardt b, Jonathan Smallwood a, Elizabeth Jefferies a,⁎
a
Department of Psychology and York Neuroimaging Centre, University of York, UK
b
McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital, McGill University, Montreal, QC H3A 2B4, Canada
a r t i c l e i n f o a b s t r a c t
Article history: Making sense of the world around us depends upon selectively retrieving information relevant to our current
Received 25 February 2016 goal or context. However, it is unclear whether selective semantic retrieval relies exclusively on general control
Revised 26 April 2016 mechanisms recruited in demanding non-semantic tasks, or instead on systems specialised for the control of
Accepted 21 May 2016
meaning. One hypothesis is that the left posterior middle temporal gyrus (pMTG) is important in the controlled
Available online 25 May 2016
retrieval of semantic (not non-semantic) information; however this view remains controversial since a parallel
Keywords:
literature links this site to event and relational semantics. In a functional neuroimaging study, we demonstrated
Semantic control that an area of pMTG implicated in semantic control by a recent meta-analysis was activated in a conjunction of
Memory retrieval (i) semantic association over size judgements and (ii) action over colour feature matching. Under these circum-
Default mode network stances the same region showed functional coupling with the inferior frontal gyrus — another crucial site for se-
Multidemand network mantic control. Structural and functional connectivity analyses demonstrated that this site is at the nexus of
Posterior middle temporal gyrus networks recruited in automatic semantic processing (the default mode network) and executively demanding
tasks (the multiple-demand network). Moreover, in both task and task-free contexts, pMTG exhibited functional
properties that were more similar to ventral parts of inferior frontal cortex, implicated in controlled semantic re-
trieval, than more dorsal inferior frontal sulcus, implicated in domain-general control. Finally, the pMTG region
was functionally correlated at rest with other regions implicated in control-demanding semantic tasks, including
inferior frontal gyrus and intraparietal sulcus. We suggest that pMTG may play a crucial role within a large-scale
network that allows the integration of automatic retrieval in the default mode network with executively-
demanding goal-oriented cognition, and that this could support our ability to understand actions and non-
dominant semantic associations, allowing semantic retrieval to be ‘shaped’ to suit a task or context.
© 2016 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
Introduction to recover weak associations (carrot-horse) and to match words on

the basis of specific sensory-motor features, such as actions or colour
Across our lifetime we acquire a large body of conceptual knowl- (e.g., carrot with traffic cone), since the functional characteristics of
edge, only a subset of which is relevant for any given task or context; these concepts are more central to their meaning (Thompson-Schill
thus automatic spreading activation within semantic representations et al., 1997; Badre et al., 2005; Whitney et al., 2011; Noonan et al.,
is often insufficient for efficient semantic cognition (Thompson-Schill 2013; Davey et al., 2015a).
et al., 1997; Badre et al., 2005; Jefferies, 2013). Automatic spreading ac- Different brain regions have been implicated in the representation
tivation can facilitate the retrieval of features and associations that are and controlled retrieval of semantic information. The ventral anterior
dominant for a particular concept (e.g., carrot-peel). When semantic re- temporal lobes (ATLs) have been argued to form a key repository of con-
trieval needs to be focussed on aspects of knowledge that are not the ceptual information, following studies of patients with semantic de-
strongest response for the inputs, additional control mechanisms can mentia (SD). These patients have relatively focal bilateral atrophy
be engaged to guide semantic retrieval. For example, control is needed focussed on ATL, associated with a gradual deterioration of knowledge
and multimodal semantic deficits, first affecting fine-grained distinc-
⁎ Corresponding author at: Department of Psychology, University of York, York YO10
tions between concepts, and then eroding more basic distinctions
5DD, UK. (Mummery et al., 2000; Hodges and Patterson, 2007; Patterson et al.,
E-mail address: [email protected] (E. Jefferies). 2007). Deficits in SD patients suggest they show a loss of central
http://dx.doi.org/10.1016/j.neuroimage.2016.05.051
1053-8119/© 2016 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
166 J. Davey et al. / NeuroImage 137 (2016) 165–177
semantic information (Bozeat et al., 2000; Jefferies and Lambon Ralph, Fig. 1 illustrates the spatial distribution for these three putative net-
2006) and studies employing inhibitory transcranial magnetic stimula- works (MDN, DMN, and semantic control) from prior published investi-
tion (TMS) in healthy participants have provided converging evidence gations. This figure shows that regions implicated in semantic control by
for a necessary role of this region in comprehension (Pobric et al., the meta-analysis of Noonan et al. (2013, in green) are only partially
2007, 2010). Functional magnetic resonance imaging (fMRI) studies re- overlapping with the MDN (from Fedorenko et al., 2013, in red). Non-
veal activation of ATL during diverse semantic judgements (Binder et al., overlapping areas in LIFG and pMTG appear to be important for de-
2009; Visser et al., 2010; Rice et al., 2015). Finally, analyses of inter- manding semantic tasks (relative to easier semantic judgements) but
regional signal correlations during task free (i.e. resting-state) functional not executive control across domains. Moreover, these semantic control
scans have shown that ATL is part of a large scale assembly that includes regions are spatially intermediate between the MDN (implicated in ex-
medial prefrontal and posterior cingulate cortices, commonly referred to ecutive control) and the DMN (implicated in automatic retrieval, from
as the default mode network (DMN, Raichle et al., 2001; Buckner et al., Yeo et al., 2011, in blue); this location could allow semantic control re-
2008; Yeo et al., 2011; Jackson et al., 2016). gions to integrate two distributed networks that are anti-correlated at
Converging neuroscientific methods have also identified brain re- rest and yet both crucial for semantic cognition, e.g., when semantic
gions beyond ATL which are important for multimodal semantics, spe- knowledge, not a task goal, defines the attentional focus.
cifically left inferior frontal gyrus (LIFG) and posterior middle The proposal that the control of semantic retrieval is partially dis-
temporal gyrus (pMTG). These regions are thought to contribute to tinct from executive control is broadly consistent with functional disso-
the control of semantic retrieval. Patients with semantic aphasia (SA), ciations that have been identified within left inferior frontal cortex.
who have lesions affecting these regions following stroke, fail the Within the language domain, studies have reported a functional gradi-
same range of verbal and non-verbal semantic tasks as SD patients; ent in left inferior frontal gyrus (IFG), with ventral anterior aspects of
however, unlike SD cases, they often retrieve information that is irrele- IFG implicated in semantic control specifically, and dorsal posterior
vant or inappropriate for the task, show strong effects of cues and mis- IFG contributing more broadly to language control, including phonolog-
cues, and perform poorly in the face of strong distracters or ambiguous ical tasks (Poldrack et al., 1999; Wagner et al., 2001a, 2001b; Devlin
meanings (Thompson-Schill et al., 2002; Jefferies and Lambon Ralph, et al., 2003; Gough et al., 2005; Snyder et al., 2007). Dorsal IFG, border-
2006; Jefferies et al., 2008, 2010; Corbett et al., 2009). Converging evi- ing inferior frontal sulcus (IFS), is recruited when participants select
dence from fMRI (Poldrack et al., 1999; Badre et al., 2005; Snijders et al., specific aspects of knowledge in line with an externally-specified goal
2010; Noonan et al., 2013; Davey et al., 2015b) and TMS (Hoffman (i.e., instructions to match words by colour or shape in the absence of
et al., 2010; Whitney et al., 2011; Davey et al., 2015a) supports the view a global semantic relationship; Badre et al., 2005). This selection process
that both of these regions contribute to semantic control. Indeed, in a may be important for many language tasks, such as lexical and phono-
recent neuroimaging meta-analysis, LIFG and pMTG were the sites acti- logical retrieval. In contrast, ventral/anterior IFG shows an increased re-
vated most strongly and consistently across many different contrasts sponse when weak and strong semantic associations are contrasted
designed to tap semantic control (Noonan et al., 2013). In addition, (e.g., salt-grain N salt-pepper) — i.e., when participants shape retrieval
when high-control semantic tasks were contrasted with demanding pho- to converge on a distant link between two concepts in the absence of
nological judgements, pMTG and the anterior part of LIFG showed a spe- an explicit goal. This ability to recover a non-dominant conceptual link
cifically semantic response, suggesting that these two regions lie outside does not generalise easily to other aspects of language processing. Re-
of the multiple-demand network (MDN), which is recruited during cent work using single-subject analyses identified regions within the
executively-demanding tasks across domains (Duncan, 2010). multiple-demand network, in dorsal and posterior IFG/IFS, that respond
These findings therefore provide some evidence that semantic cog- to difficult verbal working memory judgements involving non-words
nition may be underpinned by at least three component processes, sup- (Fedorenko et al., 2013): these regions are adjacent to, but spatially dis-
ported by distinct brain networks. (1) Domain-general executive tinct from, areas of IFG that show a greater response to easier meaning-
control implemented by the MDN (Duncan, 2010) and the fronto- based trials involving words in sentences (Fedorenko et al., 2012; Blank
parietal control system (Power and Petersen, 2013) may support the ca- et al., 2014). Moreover, analyses of resting-state connectivity have im-
pacity to engage and sustain a particular type of semantic retrieval in plicated anterior aspects of prefrontal cortex in a cingulo-opercular con-
line with the task instructions, as well as the application of top-down trol system, which includes regions that display sustained activity
constraints to support goal-driven aspects of cognition beyond seman- during task-set maintenance, while dorsal prefrontal regions couple
tics (Duncan and Owen, 2000; Duncan, 2010; Fedorenko et al., 2013; with a fronto-parietal system engaged by ongoing selection and imple-
Noonan et al., 2013). For example, in a feature-matching task (in mentation (Power and Petersen, 2013): this pattern may relate to the
which globally unrelated words must be linked together on the basis functional distinction between anterior and dorsal LIFG. Thus, a more
that they both have a particular feature specified in the task instruc- semantic response in anterior/ventral parts of IFG may be broadly in
tions), there is a need to apply a pre-specified goal during semantic re- line with the proposal that anterior areas in IFG establish and maintain
trieval, and the implementation of this goal may involve the executive priorities for what is to be retrieved, while the short-term process of se-
system. (2) Activation is thought to spread automatically between lection itself is implemented in posterior regions of IFG (Badre and
highly-related concepts within the representational system (underpin- D'esposito, 2007). Badre and colleagues referred to this functional spe-
ning semantic priming effects for strong associates). This allows domi- cialisation within IFG as “controlled retrieval” and “selection” respec-
nant features and associations to be retrieved in the absence of tively (Badre et al., 2005).
executive control, and is supported by ATL and potentially other regions The functional contribution of IFG has been considered in detail
in the DMN (Wirth et al., 2011; Lau et al., 2013; Power and Petersen, while the significance of the second region identified by Noonan and
2013; Jackson et al., 2016). (3) A third network might support situations colleagues, pMTG, remains controversial. Although this site is implicat-
in which there is no explicit goal to indicate which aspect of knowledge ed in semantic control, a parallel literature links pMTG, together with
should be brought to the fore, but the pattern of retrieval that is required angular gyrus (AG), to the comprehension of actions and events
for success is not the dominant one given the stimuli — i.e., semantic re- (Johnson-Frey et al., 2005; Liljeström et al., 2008), and to relational se-
trieval must be controlled to identify and sustain a linking context. The mantics (Humphreys and Lambon Ralph, 2014; Price et al., 2015), and
retrieval of relatively weak global associations is a good example of such these adjacent areas of temporoparietal cortex can show a similar re-
a task: here, the instructions do not establish which types of associations sponse to contrasts tapping event knowledge (Wagner et al., 2005;
or features should be the focus for retrieval — instead, it is necessary to Sachs et al., 2008; Kim, 2011). One theoretical account suggests
establish a linking context from the concepts themselves and retrieve that AG and/or pMTG provide a “thematic hub”, capturing aspects of
features relevant to this context. knowledge relating to the associations between concepts — such as
J. Davey et al. / NeuroImage 137 (2016) 165–177 167
Fig. 1. Spatial maps of the Default Mode Network (DMN, blue, from Yeo et al., 2011), Multiple-Demand Network (MDN, red, from Fedorenko et al., 2013) and Semantic Control Network
(green, from Noonan et al., 2013), presented on a rendered MNI-152 brain and on axial, coronal, and sagittal slices. The key for overlapping areas between different networks is presented
on the right hand side of the figure. Images are shown with fully saturated colours to maximise the visibility of the overlapping regions. Regions implicated in semantic control and also
found in the MDN include dlPFC (dorsolateral prefrontal cortex), dIFG (dorsal inferior frontal gyrus), pre-SMA (pre-supplementary motor area), IPS (intraparietal sulcus) and LOC (lateral
occipital cortex). Regions implicated in semantic control and also found in the DMN include vIFG (ventral inferior frontal gyrus); vMPFC (ventral medial prefrontal cortex) and pMTG
(posterior middle temporal gyrus).
knowledge about which concepts are found and used together functional gradient in IFG (Badre et al., 2005) by examining whether
(Schwartz et al., 2011). However, while pMTG often shows increased ventral/anterior as opposed to dorsal/posterior IFG had greater resting
activation in harder semantic tasks (Noonan et al., 2013), mid-AG typi- state connectivity with this region.
cally shows deactivation in semantic and other tasks relative to rest, es-
pecially for harder judgements (Binder et al., 2008; Humphreys and
Lambon Ralph, 2014; Humphreys et al., 2015). Moreover, these sites Method
showed a double dissociation in a recent TMS study (Davey et al.,
2015a): inhibitory stimulation of pMTG disrupted weak more than Participants
strong associations, while TMS to AG showed the opposite pattern.
These data are not easily reconciled with a simple “thematic hub” ac- This research was approved by the ethics committee of the York
count and suggest instead that pMTG and AG support different compo- Neuroimaging Centre, University of York, UK. All participants were
nents of semantic cognition, although ones that can at times function in right-handed, native English speaking, with normal or corrected-to-
a cooperative manner. normal vision. For the experimental task, we recruited 22 neurologically
The current study focussed on understanding the functional contri- healthy participants from the University of York (Cohort 1). Two partic-
bution of pMTG to semantic control and event/relational semantics: ipants were removed due to movement artefacts during fMRI data ac-
we explored the hypothesis that this region acts to integrate informa- quisition (mean age = 24.8, SD = 3.8, range 21–35 years, 9 males).
tion from the MDN and also the DMN, which are anti-correlated at Resting-state scans were collected at York for two cohorts (mean
rest. First, we identified whether there are regions of pMTG that show age = 21.3, SD = 2.7, range 18–31 years, 38 males); 39 participants
a common response to the retrieval of action features (contrasted with were recruited into Cohort 2, with 48 recruited into Cohort 3. These
colour feature judgements which have not been historically linked to two samples were collected as part of different projects; however the
pMTG; Badre et al., 2005) and global associations (relative to feature resting-state scan was collected before task-based scans in both
judgements). Both of these event/relational contrasts depend on re- cases and the data sets are combined in the analysis below. For Cohort
trieving information in line with a specific stimulus-driven context, as 3, we also obtained diffusion MRI. Finally, we also used two publicly-
opposed to the application of a specific goal specified in the instructions. available data sets to provide independent confirmation of the resting-
We compared the location of this response to regions implicated in se- state connectivity patterns observed in this study: (i) 141 participants
mantic control and domain-general control in previous meta-analyses from the Nathan Kline Institute (NKI)/Rockland Enhanced Sample
and also used psychophysiological interaction (PPI) analysis to under- (Nooner et al., 2012) were used to relate the connectivity patterns of
stand the functional coupling of this region under these conditions. Sec- ventral/anterior and dorsal/posterior IFG to the response we observed
ond, to explore whether pMTG could integrate executive and automatic in pMTG. Full details of this sample can be found in Gorgolewski et al.
mechanisms contributing to semantic retrieval, we used resting state (2014). (ii) Data from Yeo et al. (2011), implemented in Neurosynth
fMRI and diffusion MRI tractography to examine if the spatial networks (Yarkoni, 2011), was used in a final step to investigate the functional
that corresponded to peaks in our easy and hard semantic decisions connectivity of the pMTG site we obtained across different analyses.
(i.e., the contrasts of relatively global associations over the harder We compared the functional behaviour of the pMTG across these differ-
feature selection and vice versa), converged on the region of cortex ent cohorts to minimise the duration of specific testing sessions and to
identified as important in event/relational semantics. Third, we consid- allow us to capitalize on the power of large scale publicly available
ered whether the response in pMTG could be linked to the proposed data sets.
Study design brain regions responding differentially to control demands (and the re-
verse contrast for more automatic spreading activation).
Semantic knowledge for items from two semantic categories The experiment was organised into a total of 36 blocks divided
(animals and tools) was probed using three tasks (see Fig. 2). (i) The equally among the 6 experimental conditions (i.e., the 3 tasks probed
first task (global associations) involved matching probe words to a using 2 categories). There were 5 trials per block resulting in 30 trials
semantically-related target (e.g., selecting HONEYCOMB for the probe BEE, per experimental condition. Before each block commenced, an instruc-
as opposed to an unrelated distracter). This task did not require partic- tion slide was presented stating the task to be performed (global, size,
ipants to apply a specific goal or instruction to constrain semantic reaction, or colour) for 1000 ms. A reminder of the instructions was also
trieval; instead it was necessary to identify a semantic link from the present on each trial in parentheses under the probe word. A two-
items presented on each trial. (ii) In the second task, participants were alternative forced choice paradigm (Fig. 2) was used; participants
asked to identify a target which had similar dimensions to the probe were instructed to match the centrally presented probe word to one
concept (size matching) (e.g., selecting FLANNEL for the probe SANDPAPER, of two potential targets. Probe words were presented for 1000 ms,
as these items are similar in size/shape, even though they are not glob- followed by the response options which remained on screen until a re-
ally related). (iii) The final tasks required participants to match items sponse was recorded via button press, with maximum trial duration set
based on highly specific features (specific feature matching). For tool to 4500 ms. The inter-trial interval was 4000–6000 ms, with 10 s of rest
items, participants were asked to select the target word that had a sim- between each experimental block. One null event was present in each
ilar action to the probe word (e.g., selecting SCREWDRIVER for the probe experimental block to increase the amount of rest which was used as
KEY,
as these tools involve similar turning actions). For animals, partici- a baseline in the analysis of the fMRI data; the screen was blank for
pants matched items on the basis of colour similarity (e.g., selecting BAS- 4500 ms plus jitter (4000–6000 ms) with the location of the null
KETBALL for the probe TIGER, as both are orange and black). Both tasks ii event randomised in each experimental block. Before the fMRI experi-
ment, participants were given a practice session consisting of two blocks
and iii required participants to match items based on a feature given
to them in the task instructions, as opposed to identifying a link from for each condition. The task was presented in the scanner using NBS Pre-
sentation version 16. Participants viewed the words via a front silvered
the stimuli themselves. However, the specific feature matching tasks,
based on colour and action, were harder (see behavioural data below). mirror and responded using a Lumina Response Pad (Cedrus Corpora-
tion), placed in their left hand.
Given our research question, we focussed on the following contrasts:
first, the conjunction of action N colour features (localising voxels that
respond to action understanding) and global associations N size features Stimuli
(localising voxels that respond to relational judgements), expected to
converge in pMTG. By looking at the conjunction of these two sets of A copy of the stimuli used in this experiment is provided using the
contrasts, we can rule out task difficulty as a confound in our localisation Open Science Framework (OSF, https://osf.io/); https://osf.io/5pq8z/.
of the pMTG (since the global association task was easier than size All words used in the experiment were concrete nouns denoting manip-
matching). Secondly, we contrasted the hardest feature selection tasks ulable objects or animals. There were 30 animal probes and 30 man-
(action and colour matching) with easier global associations, to identify made probes, repeated across the three tasks (global association, size
Fig. 2. Example trial structure for all conditions in the task-based fMRI study. The study employed a 2 × 3 design, with three types of judgements (about global semantic associations, size
feature matching and specific feature matching) for animal and tool concepts.
matching and specific-feature matching), each with a unique target resting-state fMRI for Cohort 2 (TR = 2000 ms, TE = minimum full,
word for each task. The distracters in all conditions were target items 32 slices with 0.5 mm gap, voxel size = 3 × 3 × 3 mm3, 7 min) and
from other trials that did not have overlapping features or a global asso- resting-state fMRI for Cohort 3 (TR = 3000 ms, TE = minimum full,
ciation with the probe. No restrictions were placed on the number of 60 slices, voxel size = 3 × 3 × 3 mm3, 9 min). An intermediary FLAIR
times a word could be used (mean number of repetitions = 2, SD = scan with the same orientation as the functional scans was collected
1.4, range = 6); however the number of repetitions was equivalent to improve the co-registration between subject-specific structural and
between conditions. For the words in each trial (probe, target, and functional scans. In Cohort 3, we also collected diffusion weighted MRI
distracter) we collected measures of familiarity, imageability, manipula- data using a 2D single-shot pulsed gradient spin-echo EPI sequence
bility, lexical frequency, word length, and number of words, averaging (TR = 15,000 ms, TE = 86 ms, matrix = 96 × 96, 59 slices, voxel
across all the words in a single trial, and compared trials across relevant size = 2 × 2 × 2 mm3; b = 1000 s/mm2, 45 diffusion directions, 7 B0
conditions. Ratings of familiarity and imageability were taken from the volumes, 13 min). Parameters of the independent (NKI)/Rockland En-
MRC psycholinguistic database (Wilson, 1987). Ratings of manipulabil- hanced Sample are described in detail by Gorgolewski et al. (2014)
ity, familiarity, and imageability were also collected on a 7 point scale (1 and Smallwood et al. (2016).
— low, 7 — high) from a separate cohort of 11 healthy adult participants
who did not take part in the scanning sessions (familiarity and Data pre-processing and analysis
imageability ratings were collected for targets with missing values in
existing databases). Lexical frequency was taken from the SUBLEX-UK Task-based fMRI
database (van Heuven et al., 2014). Table 1 contains the psycholinguis- Analyses were conducted at the first and higher level using FSL-FEAT
tic variables for the experimental conditions. Trials were matched version 4.1.9 (Smith et al., 2004; Woolrich et al., 2009; Jenkinson et al.,
across conditions for word length (action vs. colour: t(58) = 1.1, p = 2012). Pre-processing included slice timing correction, linear motion
.276; global vs. size:, t(58) = 1.89, p = .076; global vs. feature: correction (Jenkinson et al., 2002), high-pass temporal filtering
t(58) = 1.46, p = .179). They were also matched for number of letters (sigma = 100 s), brain extraction (Smith, 2002), linear co-registration
(action vs. colour: t(58) = .14, p = .886; global vs. size: t(58) = .23, to the corresponding T1-weighted image followed by linear co-
p = .818; global vs. feature: t(58) = .75, p = .449), and lexical frequen- registration to MNI152 standard space (Jenkinson and Smith, 2001),
cy (action vs. colour: t(58) = 1.48, p = .114; global vs. size: t(58) = spatial smoothing using a Gaussian kernel with full-width-half-
1.13, p = .261; global vs. feature: t(58) = 1.43, p = .156). Manipulabil- maximum (FWHM) of 5 mm and grand-mean intensity normalisation
ity ratings were higher for action than colour trials as expected of the entire 4D data set by a single multiplicative factor.
(t(58) = 9.47, p b .001). Imageability was also higher for colour than ac- Pre-processed time series data were modelled using a general linear
tion trials (t(58) = 7.61, p b .001). No significant differences were ob- model correcting for local autocorrelation (Woolrich et al., 2001) using
served for manipulability and imageability for global vs. size a block design. The linear model included the six experimental condi-
(manipulability: t(58) = .027, p = .979; imageability: t(58) = 1.55, tions modelling block start time and block duration. fMRI scanning
p = .123), or for global vs. feature (manipulability: t(58) = .50, p = was split into two separate scanner runs collected sequentially; both
.616; imageability: t(58) = .322, p = .748). Finally, a different set of runs were analysed independently at the lower level then combined
13 participants rated the extent to which they found it necessary to gen- using a fixed-effects higher-level analysis. Six contrasts were defined;
erate a spatiotemporal context to complete the specific feature individual conditions N rest (animal/tool global, animal/tool size, tool
matching conditions (e.g., colour and action judgements), on a seven action, animal colour). We focussed our subsequent analysis on the
point scale (1 — not very useful, 7 — retrieving the context was very comparison of easy global associations vs. harder specific feature selec-
helpful). Retrieval of a spatiotemporal context was significantly more tion (building on the approach of Badre et al., 2005) and on a conjunc-
important for tool action trials (M = 4.13, SD = .50) than animal colour tion of action N colour and global associations N size to identify regions
trials (M = 2.49, SD = .43; t(28) = 13.87 p b .001). engaged by event/relational semantics (Nichols et al., 2005). All analy-
ses were cluster corrected using a z-statistic threshold of 2.3 to define
MRI acquisition contiguous clusters. Multiple comparisons were controlled using Gauss-
ian Random Field Theory at a threshold of p b .05.
Structural and functional data were acquired using a 3T GE HDx Ex-
cite MRI scanner utilising an eight-channel phased array head coil (GE) Psychophysiological interaction (PPI)
tuned to 127.4 MHz, at the York Neuroimaging Centre, University of A conjunction analysis of action N colour and global N size revealed
York. Structural MRI acquisition in all participants was based on a T1- an area of pMTG that responded to event semantics. We used this
weighted 3D fast spoiled gradient echo sequence (TR = 7.8 ms, TE = pMTG region as a mask and extracted the time-course (for each partic-
minimum full, flip angle 20°, matrix size = 256 × 256, 176 slices, ipant and each run) within this area to examine psychophysiological in-
voxel size = 1.13 × 1.13 × 1 mm3). Task-based and resting-state activity teractions (PPI; O'Reilly et al., 2012) between the pMTG and other brain
was recorded from the whole brain using single-shot 2D gradient-echo regions involved in event semantics. The extracted time-course of
echo planar imaging (EPI) with a flip angle = 90°, matrix size = pMTG and the interaction were included in a GLM model as explanatory
64 × 64, and field of view (FOV) = 192 × 192 mm2. Other scan param- variables (at the lower level, for each participant and each task individ-
eters slightly varied for task-based fMRI in Cohort 1 (TR = 2000 ms, ually, for each run). As with the functional analysis, the two runs were
TE = 30 ms, 32 slices, voxel size = 3 × 3 × 4.5 mm 3, 12 min), combined, and the results were submitted to a group level analysis,
Table 1
Psycholinguistic variables for the stimuli used in the task-based fMRI study.
Condition Word length Number of Manipulability Lexical Familiarity Imageability

words frequency
Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD
Global association 6.92 1.71 1.23 .26 4.77 .61 3.98 .39 5.74 .57 5.96 .51
Size feature 7.45 1.52 1.24 .27 4.77 .53 3.89 .43 5.88 .55 6.09 .42
Specific feature: action 7.13 1.64 1.27 .37 4.35 .44 3.94 .06 5.93 .46 6.30 .24
Specific feature: colour 7.63 1.87 1.28 .22 5.29 .33 3.81 .37 5.75 .49 5.67 .39
with the same cluster-forming threshold and significance level (Z = 2.3, different aspects of semantic control taken from Badre et al. (2005),
p b .05). The contrasts included in this analysis were action N colour and allowing us to link the response in pMTG to the previously-reported
global N size as before — and, as with the functional data, a formal con- functional distinction between “selection” in dorsal/posterior IFG (MNI
junction of these contrasts was conducted. − 48 18 18) and “controlled retrieval” in ventral/anterior IFG (MNI
−51 27 3).
Resting-state fMRI We also performed an ROI analysis of the task-based fMRI data using
Pre-processing steps were as for task fMRI, apart from the addition of 8 mm spheres, focused on these ventral and dorsal LIFG peaks from
Gaussian low pass temporal filtering, with sigma = 2.8 s, and spatial Badre et al. (2005) and the pMTG peak for semantic control taken
smoothing using a Gaussian kernel with full-width-half-maximum from the Noonan et al. (2013) meta-analysis (− 58 − 49 − 8). The
(FWHM) of 6 mm. We extracted the time series from 3 mm spheres FEATquery tool in FSL was used to extract average percentage signal
placed at regions of interest (ROIs, see below) and used these as explan- change across all the voxels in each ROI for all six conditions across
atory variables in connectivity analyses at the single subject level. In participants.
each analysis, we entered 11 nuisance regressors; the top five principal Following Simmons et al. (2011) we report the design choices that
components extracted from white matter (WM) and cerebrospinal fluid our study depends on. The sample size of 22 for the functional data (Co-
(CSF) masks based on the CompCor method (Behzadi et al., 2007) and hort 1) was based on the assumption that approximately 20 partici-
six head motion parameters. WM and CSF masks were generated from pants with useable data would be necessary to provide a stable
each individual's high resolution structural image (Zhang et al., 2001). measurement of the semantic processes in question. We used samples
No global signal regression was performed, following the method im- of approximately 50 participants for the diffusion MRI and 90 partici-
plemented in Murphy et al. (2009). At the group-level, analyses were pants for the resting state analysis (Cohorts 2 and 3) reflecting the
carried out using FMRIB's Local Analysis of Mixed Effects (FLAME1), data that was available; moreover, prior studies conducted in our
the same cluster correction method used for the functional fMRI was laboratory that have successfully revealed positive results with samples
used at the group level. that range from 40–90 participants (e.g. Smallwood et al., 2013, 2016).
For the NKI data, we used the same participants as in a previous
Diffusion MRI investigation (Gorgolewski et al., 2014), since the data was already
Subject-wise diffusion MRI processing was carried out in native dif- available. We did not perform a formal power calculation for any of
fusion space using FSL (version 4.1.9). Pre-processing of the DTI data in- these decisions.
volved eddy-current distortion correction and motion correction using The participants in Cohorts 2 and 3 who provided resting state and
FDT v2.0 (part of FSL), as well as brain extraction using BET. A probabi- diffusion MRI subsequently performed a behavioural battery of tasks
listic diffusion model was then fitted on the corrected data using in the laboratory. These measures were not directly related to the cur-
BEDPOSTX: the Bayesian estimation of diffusion parameters obtained rent experimental question and were not explored in the current
using sampling techniques toolbox (Behrens et al., 2003). BEDPOSTX study. The relationship between these measures and individual varia-
uses Monte Carlo Markov chain sampling to generate parameters for tion in cognitive performance is an ongoing focus in our laboratory
probabilistic tractography. Up to 2 fibres were modelled per voxel (for examples in the public record see Baker et al., 2015; Konishi et al.,
using a burn-in of 1000 iterations before starting the sampling of diffu- 2015; Smallwood et al., 2016).
sion parameters. Next, probabilistic tractography was performed to re-
construct fibres passing through our seed masks using PROBTRACKX. Results
This technique repeatedly samples from the diffusion parameters calcu-
lated in BEDPOSTX to build a distribution of the likely tracts from each Behavioural results
seed region. The seed masks were transformed from MNI standard
space to diffusion space using nonlinear registration. 5000 sample tracts Behavioural performance (reaction time, accuracy and response effi-
were generated per seed voxel. We used the standard parameters of a ciency) is shown in Table 2. A 2 (category; animals vs. tools) by 3 (task;
curvature threshold of 0.2 (corresponding to a minimum angle of ap- global association, size feature, and specific feature) repeated-measures
proximately ±80°), a step length of 0.5 mm and a maximum number analysis of variance (ANOVA) was conducted on response efficiency, re-
of steps of 2000. No waypoint or termination masks were included. vealing no significant differences between item categories (F (1,29) =
The resulting individual maps were transformed back to MNI standard 1.33, p = .258), and a significant main effect of task (F (2,58) = 19.28,
space, thresholded at 0.02% of total samples sent from the mask and p b .001), demonstrating poorest performance in the specific feature
concatenated into a single 4D file. Nonparametric voxelwise statistical condition (M = 2313), followed by the size feature (M = 2088), and
testing was performed using FSL Randomize with 25,000 permutations global association (M = 1655) matching conditions. No significant in-
in order to get a group tractography map (Nichols and Holmes, 2002). teraction was observed (F (2,58) = .42, p = .662). This pattern of results
The resulting maps were thresholded at p b 0.01, Family-Wise Error justifies the comparison of specific feature matching vs. global associa-
(FWE) corrected, using the Threshold-Free Cluster Enhancement tions as a way of identifying regions responding to difficult judgements
(TFCE) technique (Smith and Nichols, 2009). (cf. Badre et al., 2005).
Selection of seeds and ROIs Neuroimaging results
For the psychophysiological interaction (PPI) analysis we used the The unthresholded statistical maps can be found on Neurovault;
cluster generated by the conjunction of action N colour and global N http://neurovault.org/collections/WLSYBFYI/. This collection contains
size mask. For the resting state analysis we used two functional peaks the uncorrected z-statistic maps for the 6 experimental conditions (an-
from our task-based fMRI analyses (data from Cohort 1) as seeds in imal/tool global, animal/tool size, animal colour, and tool action)
the analyses of resting state connectivity and diffusion MRI (data from contrasted against rest.
Cohorts 2 and 3): one functional peak was linked to relatively automatic We identified the region of pMTG important for event/relational se-
semantic retrieval (easy global associations N hard feature selection, in mantics through a conjunction of two contrasts that commonly in-
inferior ATL, MNI co-ordinates −48 2 −38) and one was linked to exec- volved generating a spatiotemporal or thematic context to identify a
utive control (hard feature selection N easy global associations, in IFS, link between items: (i) Global semantic associations (i.e., whether
MNI co-ordinates − 42 28 16). In a further resting-state connectivity CHICKEN goes with EGG) were compared with feature decisions about ob-
analysis (using the NKI data), we placed seeds at peaks responding to ject size (i.e., whether a TORTOISE is the same size as a HELMET), since global
Table 2
Behavioural results (RT, accuracy, and response efficiency).
Condition Response efficiency RT Accuracy
Mean SE Mean (milliseconds) SE Mean (% correct) SE
Animal global 1636.8 59.9 1565.1 52.5 .96 .01

Tool global 1984.2 94.2 1594.7 62.7 .97 .01
Animal size 2292.7 154.3 1830.5 52.2 .94 .01
Tool size 1674.9 84.5 1896.7 45.8 .90 .02
Animal colour 2192.8 112.2 1866.9 50.1 .86 .03
Tool action 2333.4 118.5 2031.7 61.1 .90 .02
SE = standard error.
associations to both animals and tools require a linking context to be between these contrasts revealed the hypothesised pattern of shared
recovered, while size matching does not. (ii) Decisions about action fea- activation in pMTG, indicating that this region was common to both ac-
tures (i.e., whether the motion used by a KEY is similar to a SCREWDRIVER), tion and relational semantic judgements (shown in green in Fig. 3, both
were compared with decisions about colour features (i.e., whether a top and bottom panels).
TIGER is the same colour as a BASKETBALL), since tool action judgements in- To understand if this region of pMTG was independent of regions
volved generating a spatiotemporal framework to support retrieval, that were exclusively activated by either difficult or easy tasks we com-
while animal colour judgements did not. While these contrasts are dif- pared the conjunction in Fig. 3 (shown in green), with the contrasts of
ferent in many ways, pMTG was expected to respond to both since it specific feature selection over global associations (hard N easy semantic
has been implicated in understanding actions and thematic associa- decisions, shown in red in Fig. 3, bottom panel) and global semantic as-
tions. Importantly, a response to this conjunction cannot be explained sociations over specific feature selection (easy N hard semantic deci-
in terms of global task difficulty — since the global association task sions, shown in blue in Fig. 3, bottom panel). Consistent with prior
was easier than the size judgement task, according to behavioural studies (Duncan and Owen, 2000; Fedorenko et al., 2013), hard N easy
performance. decisions engaged regions of the MDN, whereas easy N hard decisions
The result of these analyses is presented in Fig. 3. When compared to activated regions in the DMN. The conjunction of action and relational
size judgements, global associations activated posterior aspects of the judgements (in green in Fig. 3) fell between these networks suggesting
temporal lobes extending from the lateral occipital cortex, along the that the pMTG's role in processing actions and relations is consistent
middle temporal gyrus into the anterior temporal lobes. Activation with a possible role for this site in the integration of information from
was also observed in the left frontal lobe focused on middle frontal the multiple-demand network and DMN.
gyrus, superior frontal gyrus and frontal pole (contrast shown in blue Next, to assess the possibility that the pMTG is part of a network of
in top panel of Fig. 3). Relative to colour judgements, action judgements regions that is important for the functional integration of information
activated a large left temporoparietal cluster including inferior lateral from the MDN and the DMN, we conducted a psychophysiological inter-
occipital cortex, posterior middle and inferior temporal gyrus, action (PPI) analysis to characterise its functional connectivity during
supramarginal gyrus, superior parietal lobule and angular gyrus. A sec- active task processing. We used the region of the pMTG that responded
ond inferior frontal cluster revealed activation in precentral gyrus, to the conjunction of Action N Colour and Global N Size as a seed region
LIFG (opercularis and triangularis) and frontal orbital cortex (contrast and at the group level explored the pattern of functional connectivity
shown in red in top panel of Fig. 3). The formal conjunction analysis that was common to both forms of event/relational semantics. The
Fig. 3. Functional activation for the conjunction of global semantic decisions and action features (top row), and the contrast between easy and hard semantic decisions (bottom row). The
top row displays activation for global semantic associations N size feature selection (blue), and for tool action feature selection N animal colour feature selection (red). The conjunction
revealing the shared activation between these contrasts (event/relational semantics) is shown in green. The bottom row shows the contrast of hard N easy semantic decisions (specific
feature selection trials N easier global association judgements, in red) and easy N hard semantic decisions (the reverse contrast, in blue). The event semantics conjunction from the top
row is overlaid (in green). All contrasts and conjunctions are cluster corrected for multiple comparisons (inclusion threshold z = 2.3, cluster significance = p b .05).
results of this analysis are presented in Fig. 4: there was greater func- anterior/ventral IFG, and “selection” in posterior/dorsal IFG). We rea-
tional connectivity during event/relational semantics in a large region soned that if pMTG is implicated in specific aspects of controlled seman-
of inferior prefrontal cortex. Comparisons of this spatial map with the tic retrieval (e.g., the generation and application of a spatiotemporal or
DMN revealed a region of overlap in the ventral inferior frontal gyrus, thematic context to identify a link between concepts), as opposed to dif-
while a comparison with the MDN revealed overlap in more ventral re- ficult semantic judgements in general, it may show stronger connectiv-
gions of lateral prefrontal cortex. ity to anterior/ventral portions of IFG. Our prior PPI analysis yielded a
During tasks that rely on event/relational semantics, the pMTG cluster of activity that extended from the dorsal region corresponding
shows increased activation and heightened communication with re- to Badre et al.'s (2005) peaks for selection (−48 18 18) to the ventral
gions in the MDN and DMN, indicating a role for posterior regions of region corresponding to their peak for controlled retrieval (− 51 27
the temporal lobe in the integration between these two large-scale net- 3) indicating that functionally the pMTG increased communicated
works. To understand whether this integrative role arises because with both of these regions during event semantics. To understand if
pMTG is at the nexus of the DMN and MDN we performed a seed the resting profile of the pMTG was more similar to the region impor-
based resting state fMRI connectivity analysis in an independent set of tant in controlled retrieval, we calculated the differences in the func-
participants (Cohorts 2 and 3). We placed seeds around the peak re- tional connectivity profiles of the two seeds using the NKI data set.
sponse in IFS using the contrast of feature selection N global associations Regions that showed greater connectivity with anterior/ventral IFG
(to identify a region in the MDN implicated in executive control; −42 (compared with posterior/dorsal IFG) included anterior temporal
28 16). We used the reverse contrast of global associations N feature se- lobes, medial prefrontal cortex, angular gyrus and a region of pMTG
lection in the ATL (to identify a site in the DMN implicated in semantic that overlapped with our previous task-based analysis (see Fig. 6 top
representation and automatic retrieval; −48 2 −38). A formal conjunc- panel). This illustrates that although during event/relational semantics,
tion between these connectivity maps identified two regions common pMTG communicates with both dorsal and ventral regions of the inferi-
to both networks, in LIFG and pMTG (overlapping with the cluster or frontal cortex, at rest this region of posterior temporal cortex is more
found in the functional study; see Fig. 5). This spatial pattern was similar functionally coupled to ventral regions of inferior frontal cortex.
to the functional recruitment observed in the PPI analysis (see subpan- We also performed a region of interest (ROI) analysis of the task-
el). In addition, we examined the fibre tracts from our ATL and IFS seed based data to examine the correspondence between the pattern of re-
regions using diffusion tractography. Overlapping white matter (WM) sponse in anterior/ventral and posterior/dorsal IFG with the functional
tracts were observed beneath our conjunction pMTG cluster (see recruitment in pMTG, localised using the peak for semantic control
Fig. 5, right-hand column). These results suggest that the region of from the meta-analysis of Noonan et al. (2013). This analysis revealed
pMTG implicated in event/relational semantics, and in semantic control similarities between pMTG and anterior/ventral IFG — with both sites
by the wider literature (Kim, 2011; Noonan et al., 2013; Davey et al., responding more to (i) global associations vs. size feature judgements,
2015b), has a pattern of functional and structural connectivity that as well as (ii) action feature matching vs. colour feature matching (see
would allow it to coordinate networks implicated in automatic semantic Fig. 6 bottom panel). In contrast, posterior/dorsal IFG responded most
retrieval and executive control. strongly to the two specific feature selection tasks and showed no pref-
We next related the region of pMTG implicated in the analyses above erence for the global association task over size matching: thus, percent-
to functional gradients previously reported for control-demanding age signal change in this dorsal region largely mirrored the difficulty of
semantic tasks in LIFG. Our aim was to investigate whether this region the conditions.
shows a pattern of connectivity compatible with a large-scale distribut- Our final analysis directly examined the similarity between the func-
ed network underpinning a particular aspect of semantic control tional architecture of the pMTG region, identified by our analyses as im-
(i.e., Badre et al.'s, 2005 distinction between “controlled retrieval” in portant in event semantics, with the network associated with semantic
Fig. 4. The rendered image shows the results of the psychophysiological interaction (PPI) analysis for the region showing common activation for the global semantic decisions and action
features (green). The axial slices illustrate regions of overlap between this spatial map and the default mode network (DMN) and the multiple demand network (MDN). For ease of visual
comparison the grey box presents the overlap between the meta-analysis of Semantic control and the same pair of large-scale networks. The PPI was cluster corrected for multiple
comparisons (inclusion threshold z = 2.3, cluster significance = p b .05).
Fig. 5. Resting state functional connectivity and structural connectivity (tractography) for functional peaks identified for hard semantic judgements in a key executive region (−42 28 16,
inferior frontal sulcus — IFS, in red) and easy semantic judgements in a region linked to semantic representation (−48 2 −38, anterior temporal lobe — ATL, in blue). The conjunction for
the two connectivity patterns is displayed in green. All contrasts and conjunctions are cluster corrected to control for multiple comparisons (inclusion threshold z = 2.3, cluster
significance = p b .05). The left-hand column shows the seed regions, columns 2 and 3 show resting state connectivity and white matter (WM) fibre tracts identified using diffusion
MRI for each seed and their overlap are displayed on the right. We present the spatial maps from the Noonan et al. (2013) meta-analysis and from the prior PPI analysis to facilitate
visual comparison of these three networks. Note the colour bar does not refer to the DTI Images which were corrected using randomise and are presented as fully saturated maps.
control by the meta-analysis of Noonan et al. (2013). We calculated the judgements about global semantic associations and action features
conjunction of the previous three analyses (the task-based conjunction and found that under these conditions it showed greater functional cou-
for action/relational semantic decisions; the resting-state connectivity pling with inferior regions of the left frontal prefrontal cortex that in-
conjunction for ATL and IFS; and the resting-state connectivity contrast cluded aspects of both the DMN and the MDN. This region overlapped
of ventral IFG N dorsal IFG). This identified a common region centred on with an area implicated in semantic control by a recent meta-analysis
− 57 − 55 3 (Fig. 7 left panel). We entered this coordinate in (Noonan et al., 2013). Resting state functional connectivity in an
Neurosynth (Yarkoni, 2011; Yarkoni et al., 2011) to examine its independent data set revealed that the same region was intrinsically
resting-state functional connectivity. The resulting map overlapped coupled at rest to seed regions exhibiting peak activity in hard N easy se-
with the Noonan et al. (2013) meta-analysis in inferior frontal gyrus mantic judgements (in inferior frontal sulcus, IFS, within the multiple-
and also showed connectivity with temporal parietal junction and demand executive network) and easy N hard decisions (in anterior tem-
pre-SMA (Fig. 7 right panel). The connectivity map for pMTG was also poral lobe; ATL). A similar analysis using diffusion MRI data demonstrat-
compared with the DMN and the multiple-demand network (Fig. 8) ed that long-range connections from IFS and ATL overlapped in white
and it extended into both of these networks, including overlap with matter adjacent to pMTG. Together these findings show that in topolog-
the MDN in IFS, precentral gyrus (PCG), dorsal IFG, IPS and LOC, plus ical terms, pMTG is located at the intersection of the DMN and MDN, a
overlap with the DMN in ventral IFG and MTG. position that would allow it to integrate information from otherwise
anti-correlated large-scale systems.
Discussion Our findings also link the functions of pMTG to previously-reported
dissociations between different aspects of semantic control within left in-
This study set out to better characterise the functional role of the ferior frontal gyrus (LIFG; Badre et al., 2005). Anterior ventral LIFG is im-
posterior middle temporal gyrus (pMTG) in semantic processing. Left plicated in ‘controlled retrieval’, while posterior dorsal LIFG is involved in
pMTG is thought to play a key role in both controlled aspects of seman- the ‘selection’ of relevant information (Badre et al., 2005). Our study sug-
tic retrieval and the comprehension of events, relations and actions — gests that this distinction extends to the posterior temporal cortex: pMTG
but the specific relationship between structure and function remains was differentially linked to ventral IFG, consistent with previous work
an open question. Although the field of semantic cognition is converging which identified strong connectivity between these regions for semanti-
on a component process account, involving conceptual representations cally demanding sentences (Snijders et al., 2010). In contrast, posterior in-
plus control mechanisms which can shape the pattern of retrieval to suit ferior temporal gyrus (ITG) and lateral occipital cortex (LOC; below but
the task or context, the brain mechanisms that underpin this capacity adjacent to pMTG) were coupled with IFS and implicated in demanding
remain poorly understood. Our analysis suggests that pMTG is a func- feature selection tasks. Our ROI analysis of the task-based data revealed
tional nexus drawing together two well-documented large-scale net- that both ventral IFG (i.e., the site implicated in controlled retrieval by
works implicated in automatic semantic processing and executive Badre et al., 2005) and pMTG (the site linked to semantic control by
control, and thus allowing more controlled patterns of retrieval. In Noonan et al., 2013) showed a pattern of functional recruitment consis-
task-based fMRI, we identified pMTG through the conjunction of tent with a role in event comprehension — i.e., a stronger response to
Fig. 6. The upper panel shows the difference in resting-state connectivity between the dorsal (red) and ventral (green) inferior frontal gyrus (IFG) peaks from Badre et al. (2005), presented
on MNI-152 axial and coronal slices. All difference maps are cluster corrected to control for multiple comparisons (inclusion threshold z = 2.3, cluster significance = p b .05). Colour bars
represent the strength of the difference in the connectivity profiles between the two seed regions in LIFG. The lower panel shows the percent signal change for all experimental conditions
extracted from 8 mm regions of interest (ROIs) for dorsal and ventral IFG from Badre et al. (2005) and for posterior middle temporal gyrus (pMTG) from Noonan et al. (2013). For ease of
visual presentation we present the overlap between the seed regions in IFG and the resting state connectivity presented in Fig. 5 in the grey box. The black bars represent percentage signal
change for the animal conditions, and the grey bars represent the signal change for the tool conditions. Error bars correspond to the standard error, with p values for between-condition t-
tests presented below. G = global associations. S = size feature matching. F = specific feature matching.
action vs. colour judgments and to global vs. size judgements. In correspondence between event semantics and a meta-analysis of se-
contrast, dorsal IFG showed a stronger response in the hardest fea- mantic control rules out simple interpretations of this region as
ture matching trials, irrespective of the need to identify a spatiotem- supporting only one or other of these aspects of processing. Likewise,
poral or thematic context: both colour and action features elicited our demonstration that pMTG is a convergence zone for both the
more activation in this region than global semantic associations. DMN and the MDN, while occupying a spatial location that is indepen-
pMTG also showed stronger connectivity with ventral than dorsal dent of regions recruited by easy and hard decisions, indicates that
IFG in resting-state analyses. This pattern is consistent with the this region is not an exclusive member of either network. Our PPI anal-
view that dorsal IFG bordering IFS is implicated in the application ysis shows that during event semantics the pMTG becomes functionally
of a goal to semantic selection, while ventral IFG and pMTG allow re- coupled to both regions of the DMN and the MDN, providing evidence
trieval to be shaped to suit a context defined by the stimuli that the pMTG is in communication with multiple networks when se-
themselves (i.e., link is not specified by the task instructions). mantic retrieval must be shaped to fit different contexts. Contextual
Global associations, especially those tapping relatively weak, non- shaping of retrieval is common to both event semantics and semantic
dominant relationships (which elicit more activation than strong as- control, and could be made possible by the location of the pMTG as a
sociations in pMTG across multiple studies; Badre et al., 2005; Gold nexus between the DMN and the MDN. Since unguided spreading acti-
et al., 2005; Noonan et al., 2013; Krieger-Redwood et al., 2015), re- vation in ATL recovers strong associations and features, even when
quire retrieval mechanisms to determine a context which links these are irrelevant, this process may lead to problems in retrieval
these items together and applied to ensure retrieval stays focussed when the appropriate target is not the most dominant aspect of our
on the association being probed. Action feature matching also re- knowledge. By integrating information from the DMN and MDN the
quires a spatiotemporal context to be determined and used to pMTG may allow a set of features to be maintained that can help
guide the retrieval of actions. Finally, we used the conjunction of shape ongoing spreading activation to suit a particular spatiotemporal
the location of the pMTG from each of these analyses as a seed region or thematic context, as occurs in both weak associations, and tasks
in a publicly available data set, yielding a pattern of connectivity that that depend upon spatiotemporal context such as event semantics.
was consistent with a meta-analysis of studies of semantic control Our analysis highlights that pMTG is spatially well-placed to provide
(Noonan et al., 2013). this constraint on semantic retrieval since it is located between systems
Together these lines of structural and functional evidence constrain that are important for automatic semantic retrieval and top-down goal-
the possible role of pMTG in semantic cognition. The spatial directed cognition.
Fig. 7. The left hand panel shows overlapping clusters in posterior middle temporal gyrus (pMTG) across different contrasts and analyses; (i) the event semantics task-based conjunction
(green, reproduced from Fig. 3), (ii) the inferior frontal sulcus (IFS) and anterior temporal lobe (ATL) connectivity conjunction (red, reproduced from Fig. 5), and (iii) the difference
between ventral left inferior frontal gyrus (vLIFG) and dorsal LIFG (dLIFG) connectivity (blue, reproduced from Fig. 6). In the main figure, the top row displays the Neurosynth
functional connectivity pattern for a seed corresponding to the centre of gravity (COG) for the cluster where all three contrasts overlap. The bottom row compares this pattern for
pMTG (in green) with the Noonan et al. (2013) semantic control meta-analysis from Fig. 1 (in red). Regions that fall within both maps are shown in yellow. Images are shown with
fully saturated colours to maximise the visibility of the overlapping regions.
Our study is also informative regarding the functional similarities related deactivation along with aspects of ATL, especially for non-
and differences between angular gyrus (AG) and pMTG. Consistent semantic tasks and harder semantic tasks (Humphreys et al., 2015).
with the hypothesis that AG/pMTG form a convergence zone for the- While AG showed a response to easy global associations over hard fea-
matic knowledge, both of these sites have been implicated in event se- ture selection in this study, and strong resting state connectivity to ATL,
mantics and thematic associations, as opposed to object knowledge pMTG did not show the same profile — the response of this region was
(Schwartz et al., 2011). Nevertheless, pMTG has previously been impli- not explicable in terms of overall difficulty (or reverse difficulty) and it
cated alongside LIFG in semantic control, while AG tends to show task- showed connectivity at rest to both ATL and IFS. Our data, therefore, add
Fig. 8. Functional connectivity of posterior middle temporal gyrus (pMTG; in green, reproduced from Fig. 7) contrasted with the multiple-demand network (MDN; in red) and default
mode network (DMN; in blue). It can be seen that the connectivity of the pMTG intersects with the MDN and DMN in similar regions as the Noonan meta-analysis (see subpanel).
Images are shown with fully saturated colours to maximise the visibility of the overlapping regions. vIFG = ventral inferior frontal gyrus; LOC = lateral occipital cortex; IFS = inferior
frontal sulcus; dIFG/PCG = dorsal inferior frontal gyrus/precentral gyrus; IPS = intraparietal sulcus. For ease of visual comparison in the grey panel we show the spatial maps
generated through our prior three analyses, as well the Noonan meta-analysis.
to a growing body of evidence that while AG and pMTG sometimes co- propagation of uncertainty in diffusion-weighted MR imaging. Magn. Reson.
Med. 50, 1077–1088.
activate during semantic tasks, they do not always do so, and this may Behzadi, Y., Restom, K., Liau, J., Liu, T.T., 2007. A component based noise correction meth-
be in part a consequence of their differential patterns of functional od (CompCor) for BOLD and perfusion based fMRI. NeuroImage 37, 90–101.
connectivity. Binder, J.R., Swanson, S.J., Hammeke, T.A., Sabsevitz, D.S., 2008. A comparison of five
fMRI protocols for mapping speech comprehension systems. Epilepsia 49,
In summary, the current study helps constrain component process 1980–1997.
accounts of semantic cognition. There is an emerging consensus that Binder, J.R., Desai, R.H., Graves, W.W., Conant, L.L., 2009. Where is the semantic system? A
the brain regions reliably activated by semantic tasks can show different critical review and meta-analysis of 120 functional neuroimaging studies. Cereb. Cor-
tex 19, 2767–2796.
response profiles — for example, meta-analytic investigations of neuro-
Blank, I., Kanwisher, N., Fedorenko, E., 2014. A functional dissociation between language
imaging evidence, work with patient populations and TMS studies in and multiple-demand systems revealed in patterns of BOLD signal fluctuations.
normal participants suggest that brain regions implicated in automatic J. Neurophysiol. 112, 1105–1118.
Bozeat, S., Lambon Ralph, M.A., Patterson, K., Garrard, P., Hodges, J.R., 2000. Non-verbal se-
semantic retrieval and representation can be dissociated from others
mantic impairment in semantic dementia. Neuropsychologia 38, 1207–1215.
implicated in more controlled semantic processing (Jefferies and Buckner, R.L., Andrews-Hanna, J.R., Schacter, D.L., 2008. The brain's default network: anat-
Lambon Ralph, 2006; Jefferies, 2013; Noonan et al., 2013; Humphreys omy, function, and relevance to disease. Ann. N. Y. Acad. Sci. 1124, 1–38.
and Lambon Ralph, 2014). In addition, only a subset of the regions im- Corbett, F., Jefferies, E., Lambon Ralph, M.A., 2009. Exploring multimodal semantic
control impairments in semantic aphasia: evidence from naturalistic object
plicated in semantic control contributes to executively-demanding use. Neuropsychologia 47, 2721–2731.
non-semantic judgements (Noonan et al., 2013). In line with this litera- Davey, J., Cornelissen, P.L., Thompson, H.E., Sonkusare, S., Hallam, G., Smallwood, J.,
ture, our data provide converging evidence for at least three compo- Jefferies, E., 2015a. Automatic and controlled semantic retrieval: TMS reveals distinct
contributions of posterior middle temporal gyrus and angular gyrus. J. Neurosci. 35,
nents of semantic cognition. First, a network engaging ATL, AG, medial 15230–15239.
PFC and PCC, broadly corresponding to the so-called default mode net- Davey, J., Rueschemeyer, S.-A., Costigan, A., Murphy, N., Krieger-Redwood, K., Hallam, G.,
work, responds to more strongly to relatively easy global associations, Jefferies, E., 2015b. Shared neural processes support semantic control and action un-
derstanding. Brain Lang. 142, 24–35.
suggesting that it preferentially supports automatic spreading activa- Devlin, J.T., Matthews, P.M., Rushworth, M.F.S., 2003. Semantic processing in the left
tion within semantic representations. Second, a network including IFS, inferior prefrontal cortex: a combined functional magnetic resonance imaging
dorsal posterior IFG, IPS, pre-SMA and ITG/LOC, corresponding to the and transcranial magnetic stimulation study. J. Cogn. Neurosci. 15, 71–84.
Duncan, J., 2010. The multiple-demand (MD) system of the primate brain: mental pro-
multiple-demand executive network, is recruited for demanding fea-
grams for intelligent behaviour. Trends Cogn. Sci. 14, 172–179.
ture decisions suggesting that it might be important for the top down al- Duncan, J., Owen, A.M., 2000. Common regions of the human frontal lobe recruited by di-
location of effort to match current environmental demands. Finally, verse cognitive demands. Trends Neurosci. 23, 475–483.
Fedorenko, E., Duncan, J., Kanwisher, N., 2012. Language-selective and domain-general re-
pMTG and aIFG formed a third network showing common recruitment
gions lie side by side within Broca's area. Curr. Biol. 22, 2059–2062.
in situations when semantic retrieval must be shaped in a flexible way Fedorenko, E., Duncan, J., Kanwisher, N., 2013. Broad domain generality in focal regions of
to suit a spatiotemporal or thematic context (even for global associa- frontal and parietal cortex. Proc. Natl. Acad. Sci. U. S. A. 110, 16616–16621.
tions; i.e., in the absence of task instructions that require goal-driven Gold, B.T., Balota, D.A., Kirchhoff, B.A., Buckner, R.L., 2005. Common and dissociable acti-
vation patterns associated with controlled semantic and phonological processing: ev-
retrieval). Thus, we propose that pMTG facilitates integration of infor- idence from FMRI adaptation. Cereb. Cortex 15, 1438–1450.
mation from regions in the DMN (that underpin more automatic aspects Gorgolewski, K.J., Lurie, D., Urchs, S., Kipping, J.A., Craddock, R.C., Milham, M.P., Margulies,
of semantic cognition) and those in the MDN (that contribute to D.S., Smallwood, J., 2014. A correspondence between individual differences in the
brain's intrinsic functional architecture and the content and form of self-generated
executively-demanding judgements), together with more ventral as- thoughts. PLoS One 9, e97176.
pects of IFG. This third component of semantic cognition relies on the Gough, P.M., Nobre, A.C., Devlin, J.T., 2005. Dissociating linguistic processes in the left inferior
integration of DMN and MDN that would otherwise remain anti- frontal cortex with transcranial magnetic stimulation. J. Neurosci. 25, 8010–8016.
Hodges, J.R., Patterson, K., 2007. Semantic dementia: a unique clinicopathological syn-
correlated, allowing the retrieval of semantic representations to be drome. Lancet Neurol. 6, 1004–1014.
shaped to fit the current demands. Hoffman, P., Jefferies, E., Lambon Ralph, M.A., 2010. Ventrolateral prefrontal cor-
tex plays an executive regulation role in comprehension of abstract words:
convergent neuropsychological and repetitive TMS evidence. J. Neurosci. 30,
Acknowledgements 15450–15456.
Humphreys, G.F., Lambon Ralph, M.A., 2014. Fusion and fission of cognitive functions in
the human parietal cortex. Cereb. Cortex.
EJ was supported by grants from BBSRC (BB/J006963/1) and the Humphreys, G.F., Hoffman, P., Visser, M., Binney, R.J., Lambon Ralph, M.A., 2015. Establish-
European Research Council (SEMBIND — 283530). This study was made ing task- and modality-dependent dissociations between the semantic and default
possible through the support of a grant from the John Templeton Founda- mode networks. Proc. Natl. Acad. Sci. 112, 201422760.
Jackson, R.L., Hoffman, P., Pobric, G., Lambon Ralph, M.A., 2016. The semantic network at
tion to JS, “Prospective Psychology Stage 2: A Research Competition” to
work and rest: differential connectivity of anterior temporal lobe subregions.
Martin Seligman. The opinions expressed in this publication are those of J. Neurosci. 36, 1490–1501.
the author(s) and do not necessarily reflect the views of the John Jefferies, E., 2013. The neural basis of semantic cognition: converging evidence from neu-
Templeton Foundation. HET was supported by a grant part-funded by ropsychology, neuroimaging and TMS. Cortex 49, 611–625.
Jefferies, E., Lambon Ralph, M.A., 2006. Semantic impairment in stroke aphasia versus se-
the Wellcome Trust [ref: 105624] through the Centre for Chronic Diseases mantic dementia: a case-series comparison. Brain 129, 2132–2147.
and Disorders (C2D2) at the University of York. Jefferies, E., Patterson, K., Lambon Ralph, M.A., 2008. Deficits of knowledge versus execu-
tive control in semantic cognition: insights from cued naming. Neuropsychologia 46,
649–658.
Appendix A. Supplementary data Jefferies, E., Rogers, T.T., Hopper, S., Lambon Ralph, M.A., 2010. “Pre-semantic” cognition
revisited: critical differences between semantic aphasia and semantic dementia.
Neuropsychologia 48, 248–261.
Supplementary data to this article can be found online at http://dx. Jenkinson, M., Smith, S., 2001. A global optimisation method for robust affine registration
doi.org/10.1016/j.neuroimage.2016.05.051. of brain images. Med. Image Anal. 5, 143–156.
Jenkinson, M., Bannister, P., Brady, M., Smith, S., 2002. Improved optimization for the robust
and accurate linear registration and motion correction of brain images. NeuroImage 17,
References 825–841.
Jenkinson, M., Beckmann, C.F., Behrens, T.E.J., Woolrich, M.W., Smith, S.M., 2012. FSL.
Badre, D., D'esposito, M., 2007. Functional magnetic resonance imaging evidence for a hi- NeuroImage 62, 782–790.
erarchical organization of the prefrontal cortex. J. Cogn. Neurosci. 19, 2082–2099. Johnson-Frey, S.H., Newman-Norlund, R., Grafton, S.T., 2005. A distributed left hemi-
Badre, D., Poldrack, R.A., Paré-Blagoev, E.J., Insler, R.Z., Wagner, A.D., 2005. Dissociable sphere network active during planning of everyday tool use skills. Cereb. Cortex 15,
controlled retrieval and generalized selection mechanisms in ventrolateral prefrontal 681–695.
cortex. Neuron 47, 907–918. Kim, H., 2011. Differential neural activity in the recognition of old versus new events: an
Baker, D.H., Karapanagiotidis, T., Coggan, D.D., Wailes-Newson, K., Smallwood, J., 2015. activation likelihood estimation meta-analysis. Hum. Brain Mapp. 34.
Brain networks underlying bistable perception. NeuroImage 119, 229–234. Konishi, M., Mclaren, D.G., Engen, H., Smallwood, J., 2015. Shaped by the past: the default
Behrens, T.E.J., Woolrich, M.W., Jenkinson, M., Johansen-Berg, H., Nunes, R.G., Clare, mode network supports cognition that is independent of immediate perceptual
S., Matthews, P.M., Brady, J.M., Smith, S.M., 2003. Characterization and input. PLoS One 10, e0132209.
Krieger-Redwood, K., Teige, C., Davey, J., Hymers, M., Jefferies, E., 2015. Conceptual control Smallwood, J., Karapanagiotidis, T., Ruby, F., Medea, B., De Caso, I., Konishi, M., Wang, H.-
across modalities: graded specialisation for pictures and words in inferior frontal and T., Hallam, G., Margulies, D.S., Jefferies, E., 2016. Representing representation: integra-
posterior temporal cortex. Neuropsychologia. tion between the temporal lobe and the posterior cingulate influences the content
Lau, E.F., Gramfort, A., Hämäläinen, M.S., Kuperberg, G.R., 2013. Automatic semantic facil- and form of spontaneous thought. PLoS One 11, e0152272.
itation in anterior temporal cortex revealed through multimodal neuroimaging. Smith, S.M., 2002. Fast robust automated brain extraction. Hum. Brain Mapp. 17,
J. Neurosci. 33, 17174–17181. 143–155.
Liljeström, M., Tarkiainen, A., Parviainen, T., Kujala, J., Numminen, J., Hiltunen, J., Laine, M., Smith, S.M., Nichols, T.E., 2009. Threshold-free cluster enhancement: addressing prob-
Salmelin, R., 2008. Perceiving and naming actions and objects. NeuroImage 41, lems of smoothing, threshold dependence and localisation in cluster inference.
1132–1141. NeuroImage 44, 83–98.
Mummery, C.J., Patterson, K., Price, C.J., Ashburner, J., Frackowiak, R.S., Hodges, J.R., 2000. Smith, S.M., Jenkinson, M., Woolrich, M.W., Beckmann, C.F., Behrens, T.E.J., Johansen-Berg,
A voxel-based morphometry study of semantic dementia: relationship between tem- H., Bannister, P.R., De Luca, M., Drobnjak, I., Flitney, D.E., Niazy, R.K., Saunders, J.,
poral lobe atrophy and semantic memory. Ann. Neurol. 47, 36–45. Vickers, J., Zhang, Y., De Stefano, N., Brady, J.M., Matthews, P.M., 2004. Advances in
Murphy, K., Birn, R.M., Handwerker, D.A., Jones, T.B., Bandettini, P.A., 2009. The impact of functional and structural MR image analysis and implementation as FSL. NeuroImage
global signal regression on resting state correlations: are anti-correlated networks in- 23, S208–S219.
troduced? NeuroImage 44, 893–905. Snijders, T.M., Petersson, K.M., Hagoort, P., 2010. Effective connectivity of cortical and sub-
Nichols, T.E., Holmes, A.P., 2002. Nonparametric permutation tests for functional neuro- cortical regions during unification of sentence structure. NeuroImage 52, 1633–1644.
imaging: a primer with examples. Hum. Brain Mapp. 15, 1–25. Snyder, H.R., Feigenson, K., Thompson-Schill, S.L., 2007. Prefrontal cortical response to
Nichols, T., Brett, M., Andersson, J., Wager, T., Poline, J.B., 2005. Valid conjunction inference conflict during semantic and phonological tasks. J. Cogn. Neurosci. 19, 761–775.
with the minimum statistic. NeuroImage 25, 653–660. Thompson-Schill, S.L., D'Esposito, M., Aguirre, G.K., Farah, M.J., 1997. Role of left inferior
Noonan, K., Jefferies, E., Visser, M., Lambon Ralph, M.A., 2013. Going beyond inferior pre- prefrontal cortex in retrieval of semantic knowledge: a reevaluation. Proc. Natl.
frontal involvement in semantic control: evidence for the additional contribution of Acad. Sci. U. S. A. 94, 14792–14797.
dorsal angular gyrus and posterior middle temporal cortex. J. Cogn. Neurosci. 25, Thompson-Schill, S.L., Jonides, J., Marshuetz, C., Smith, E.E., D'esposito, M., Kan, I.P., Knight,
1824–1850. R.T., Swick, D., 2002. Effects of frontal lobe damage on interference effects in working
Nooner, K.B., et al., 2012. The NKI-Rockland sample: a model for accelerating the pace of memory. Cogn. Affect. Behav. Neurosci. 2, 109–120.
discovery science in psychiatry. Front. Neurosci. 6, 1–11. Van Heuven, W.J.B., Mandera, P., Keuleers, E., Brysbaert, M., 2014. SUBTLEX-UK: a new
O'Reilly, J.X., Woolrich, M.W., Behrens, T.E., Smith, S.M., Johansen-Berg, H., 2012. Tools of and improved word frequency database for British English. Q. J. Exp. Psychol. 67,
the trade: psychophysiological interactions and functional connectivity. Soc. Cogn. 1176–1190 (2006).
Affect. Neurosci. 7, 604–609. Visser, M., Embleton, K.V., Jefferies, E., Parker, G.J., Lambon Ralph, M.A., 2010. The inferior,
Patterson, K., Nestor, P.J., Rogers, T.T., 2007. Where do you know what you know? The anterior temporal lobes and semantic memory clarified: novel evidence from
representation of semantic knowledge in the human brain. Nat. Rev. Neurosci. 8, distortion-corrected fMRI. Neuropsychologia 48, 1689–1696.
976–987. Wagner, A.D., Maril, A., Bjork, R.A., Schacter, D.L., 2001a. Prefrontal contributions to exec-
Pobric, G., Jefferies, E., Lambon Ralph, M.A., 2007. Anterior temporal lobes mediate seman- utive control: fMRI evidence for functional distinctions within lateral prefrontal cor-
tic representation: mimicking semantic dementia by using rTMS in normal partici- tex. NeuroImage 14, 1337–1347.
pants. Proc. Natl. Acad. Sci. U. S. A. 104, 20137–20141. Wagner, A.D., Paré-Blagoev, E.J., Clark, J., Poldrack, R.A., 2001b. Recovering meaning: left
Pobric, G., Jefferies, E., Lambon Ralph, M.A., 2010. Category-specific versus category- prefrontal cortex guides controlled semantic retrieval. Neuron 31, 329–338.
general semantic impairment induced by transcranial magnetic stimulation. Curr. Wagner, A.D., Shannon, B.J., Kahn, I., Buckner, R.L., 2005. Parietal lobe contributions to ep-
Biol. 20, 964–968. isodic memory retrieval. Trends Cogn. Sci. 9, 445–453.
Poldrack, R.A., Wagner, A.D., Prull, M.W., Desmond, J.E., Glover, G.H., Gabrieli, J.D., 1999. Whitney, C., Kirk, M., O'sullivan, J., Lambon Ralph, M.A., Jefferies, E., 2011. The neural or-
Functional specialization for semantic and phonological processing in the left inferior ganization of semantic control: TMS evidence for a distributed network in left inferior
prefrontal cortex. NeuroImage 10, 15–35. frontal and posterior middle temporal gyrus. Cereb. Cortex 21, 1066–1075.
Power, J.D., Petersen, S.E., 2013. Control-related systems in the human brain. Curr. Opin. Wilson, M., 1987. MRC Psycholinguistic Database: Machine-usable Dictionary, Version
Neurobiol. 23, 223–228. 2.00. Science and Engineering Research Council.
Price, A.R., Bonner, M.F., Peelle, J.E., Grossman, M., 2015. Converging evidence for the neu- Wirth, M., Jann, K., Dierks, T., Federspiel, A., Wiest, R., Horn, H., 2011. Semantic memory
roanatomic basis of combinatorial semantics in the angular gyrus. 35, 3276–3284. involvement in the default mode network: a functional neuroimaging study using in-
Raichle, M.E., Macleod, A.M., Snyder, A.Z., Powers, W.J., Gusnard, D.A., Shulman, G.L., 2001. dependent component analysis. NeuroImage 54, 3057–3066.
A default mode of brain function. Proc. Natl. Acad. Sci. U. S. A. 98, 676–682. Woolrich, M.W., Ripley, B.D., Brady, M., Smith, S.M., 2001. Temporal autocorrelation in
Rice, G.E., Lambon Ralph, M.A., Hoffman, P., 2015. The Roles of Left Versus Right Anterior univariate linear modeling of FMRI data. NeuroImage 14, 1370–1386.
Temporal Lobes in Conceptual Knowledge: An ALE Meta-analysis of 97 Functional Woolrich, M.W., Jbabdi, S., Patenaude, B., Chappell, M., Makni, S., Behrens, T., Beckmann,
Neuroimaging Studies. pp. 1–18. C., Jenkinson, M., Smith, S.M., 2009. Bayesian analysis of neuroimaging data in FSL.
Sachs, O., Weis, S., Krings, T., Huber, W., Kircher, T., 2008. Categorical and thematic knowl- NeuroImage 45, S173–S186.
edge representation in the brain: neural correlates of taxonomic and thematic con- Yarkoni, T., 2011. neurosynth.orghttp://www.neurosynth.org.
ceptual relations. Neuropsychologia 46, 409–418. Yarkoni, T., Poldrack, R.A., Nichols, T.E., Van Essen, D.C., Wager, T.D., 2011. Large-scale au-
Schwartz, M.F., Kimberg, D.Y., Walker, G.M., Brecher, A., Faseyitan, O.K., Dell, G.S., Mirman, tomated synthesis of human functional neuroimaging data. Nat. Methods 8, 665–670.
D., Coslett, H.B., 2011. Neuroanatomical dissociation for taxonomic and thematic Yeo, B.T.T., Krienen, F.M., Sepulcre, J., Sabuncu, M.R., Lashkari, D., Hollinshead, M.,
knowledge in the human brain. Proc. Natl. Acad. Sci. 108, 8520–8524. Roffman, J.L., Smoller, J.W., Zöllei, L., Polimeni, J.R., Fischl, B., Liu, H., Buckner, R.L.,
Simmons, J.P., Nelson, L.D., Simonsohn, U., 2011. False-positive psychology undisclosed 2011. The organization of the human cerebral cortex estimated by intrinsic functional
flexibility in data collection and analysis allows presenting anything as significant. connectivity. J. Neurophysiol. 106, 1125–1165.
Psychol. Sci. (0956797611417632). Zhang, Y., Brady, M., Smith, S., 2001. Segmentation of brain MR images through a hidden
Smallwood, J., Gorgolewski, K.J., Golchert, J., Ruby, F.J., Engen, H.G., Baird, B., Vinski, M.T., Markov random field model and the expectation–maximization algorithm. IEEE
Schooler, J.W., Margulies, D.S., 2013. The default modes of reading: modulation of Trans. Med. Imaging 20, 45–57.
posterior cingulate and medial prefrontal cortex connectivity associated with com-
prehension and task focus while reading. Front. Hum. Neurosci. 7.

1 s2.0 S1053811916301719 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S1053811916301719 Main

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

1 s2.0 S1053811916301719 Main

Uploaded by

Copyright:

Available Formats

NeuroImage 137 (2016) 165–177

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ynimg

Exploring the role of the posterior middle temporal gyrus in semantic

Introduction to recover weak associations (carrot-horse) and to match words on

Condition Word length Number of Manipulability Lexical Familiarity Imageability

Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD

Selection of seeds and ROIs Neuroimaging results

Condition Response efﬁciency RT Accuracy

Mean SE Mean (milliseconds) SE Mean (% correct) SE

Animal global 1636.8 59.9 1565.1 52.5 .96 .01

You might also like