Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Author’s Accepted Manuscript
Interference from related actions in spoken wordproduction: Behavioural and fMRI evidence
Greig de Zubicaray, Douglas Fraser, KoriRamajoo, Katie McMahon
PII: S0028-3932(17)30010-6DOI: http://dx.doi.org/10.1016/j.neuropsychologia.2017.01.010Reference: NSY6231
To appear in: Neuropsychologia
Received date: 4 May 2016Revised date: 9 January 2017Accepted date: 9 January 2017
Cite this article as: Greig de Zubicaray, Douglas Fraser, Kori Ramajoo and KatieMcMahon, Interference from related actions in spoken word production:Behavioural and fMRI evidence, Neuropsychologia,http://dx.doi.org/10.1016/j.neuropsychologia.2017.01.010
This is a PDF file of an unedited manuscript that has been accepted forpublication. As a service to our customers we are providing this early version ofthe manuscript. The manuscript will undergo copyediting, typesetting, andreview of the resulting galley proof before it is published in its final citable form.Please note that during the production process errors may be discovered whichcould affect the content, and all legal disclaimers that apply to the journal pertain.
www.elsevier.com/locate/neuropsychologia
RUNNING HEAD: Interference from related actions 1
Interference from related actions in spoken word production: Behavioural and
fMRI evidence
Greig de Zubicaray1,*
, Douglas Fraser2, Kori Ramajoo
1, Katie McMahon
3.
1 Faculty of Health and Institute of Health and Biomedical Innovation, Queensland
University of Technology, Brisbane, Australia 2 School of Public Health, University of Queensland, Brisbane, Australia
3 Centre for Advanced Imaging, University of Queensland, Brisbane, Australia
*Corresponding author: Greig de Zubicaray Faculty of Health and Institute of Health
and Biomedical Innovation, Queensland University of Technology, Kelvin Grove,
Brisbane, QLD 4059. Australia. [email protected]
Abstract
Few investigations of lexical access in spoken word production have
investigated the cognitive and neural mechanisms involved in action naming. These
are likely to be more complex than the mechanisms involved in object naming, due to
the ways in which conceptual features of action words are represented. The present
study employed a blocked cyclic naming paradigm to examine whether related action
contexts elicit a semantic interference effect akin to that observed with categorically
related objects. Participants named pictures of intransitive actions to avoid a confound
with object processing. In Experiment 1, body-part related actions (e.g., running,
walking, skating, hopping) were named significantly slower compared to unrelated
actions (e.g., laughing, running, waving, hiding). Experiment 2 employed perfusion
functional Magnetic Resonance Imaging (fMRI) to investigate the neural mechanisms
involved in this semantic interference effect. Compared to unrelated actions, naming
related actions elicited significant perfusion signal increases in frontotemporal cortex,
including bilateral inferior frontal gyrus (IFG) and hippocampus, and decreases in
bilateral posterior temporal, occipital and parietal cortices, including intraparietal
RUNNING HEAD: Interference from related actions 2
sulcus (IPS). The findings demonstrate a role for temporoparietal cortex in
conceptual-lexical processing of intransitive action knowledge during spoken word
production, and support the proposed involvement of interference resolution and
incremental learning mechanisms in the blocked cyclic naming paradigm.
Keywords: semantic interference, action naming, fMRI
1. Introduction
A considerable body of psycholinguistic research has demonstrated that
lexical access – the process of retrieving words from the mental lexicon – can be
affected by production contexts that are similar in meaning. The majority of evidence
has come from object naming paradigms. Using these paradigms, categorically related
contexts have been demonstrated to reliably impede spoken word production
compared to unrelated contexts in both healthy participants and patients with brain
damage (e.g., Damian, Vigliocco, & Levelt., 2001; Schnur et al., 2006). The origin of
these semantic interference effects has been attributed to a conceptual preparation
stage of processing in which activation spreads between related object concepts and
their features, and subsequently to their linked lexical representations (see Belke,
2013, for a review).
Relatively few studies have investigated semantic context effects in bare
action naming. Of these, most have employed the picture-word interference (PWI)
paradigm in which target pictures are named in context with related versus unrelated
distractor words (e.g., Roelofs, 1993; Schnur, Costa, & Caramazza, 2002, Experiment
1; Vigliocco et al., 2004, Experiment 4; Vigliocco, Vinson, & Siri, 2005). As our
RUNNING HEAD: Interference from related actions 3
main interest here is in investigating analogous semantic interference effects in bare
action and object naming, we will not review studies that involved manipulations of
distractor and target grammatical class and/or required participants to name action
pictures using sentential constraints (e.g., Mahon et al., 2007; Schriefers, Teruel, &
Meinhausen, 1998). Several PWI studies have reported a semantic interference effect
in bare action naming (e.g., picture - SHAVE, distractor - comb), with the authors
concluding that similar conceptual and lexical mechanisms are likely to be engaged
for both actions and objects (e.g., Roelofs, 1993; Schnur et al., 2002; Vigliocco et al.,
2004, 2005).
Two factors complicating interpretations of semantic interference effects in
bare action naming and PWI paradigms in particular are transitivity and grammatical
ambiguity particularly when conducting experiments in English. Depictions of
transitive (i.e., object oriented) actions tend to be complex, requiring identification of
both actors and objects (even if they aren’t named) and their functional
interrelationships (including syntactic relations). For example, consider the PWI
target-distractor pairing SHAVE-comb: The depiction of the action SHAVE in the
International Picture Naming Project (IPNP) database involves an actor using a razor
in front of a mirror (Szekeley et al., 2004). Additionally, the grammatically
ambiguous distractor comb can denote both an action and an object noun categorically
related to razor. Thus, studies demonstrating semantic interference effects with
transitive actions in bare naming have likely confounded conceptual feature overlap
among object category-coordinates (e.g., Roelofs, 1993; Vigliocco et al., 2004). This
raises the question of whether “pure” semantic interference effects in intransitive
action naming are actually observable.
RUNNING HEAD: Interference from related actions 4
To address this question, spoken word production models need to specify how
action meanings are organized in the semantic system so that activation can spread
between related actions and their features, and then to their linked lexical
representations (e.g., Roelofs, 1993). The conceptual features of actions are proposed
to be primarily motoric and functional, whereas objects have a greater weighting of
sensory features (Bird, Howard, & Franklin, 2000). Words referring to actions are
therefore considered to be more abstract than words referring to objects (Vigliocco et
al., 2004). In Roelofs’ (1993) model, primitive features create an abstract conceptual-
lexical representation of the given action via a process known as chunking. In the
production system, words are accessed by using this abstract representation (see also
Vigliocco et al., 2004).
It is worth emphasizing that few production models explicitly mention actions,
and those that do fail to distinguish transitivity (e.g., Roelofs, 1993; Vigliocco et al.,
2004). According to these models, in order for intransitive actions to elicit semantic
interference in a manner analogous to categorically related objects, coactivation due
to conceptual feature overlap would need to spread to lexical competitors via a shared
superordinate category node, such as body-part relation (e.g., Abdel Rahman &
Melinger, 2009; Belke, 2013; Damian et al., 2001). In a normative study using a
body-part association task, Maouene, Hidaka and Smith (2008) showed many
individual action words learned in early childhood are systematically attributed to
specific body parts (e.g., leg/foot, hand/arm, face), while others are associated with
multiple body parts. Examples of the latter type of “whole body” action words include
swing, hide, rest and climb. Intransitive actions might therefore produce semantic
RUNNING HEAD: Interference from related actions 5
interference if body part representations are organized along the lines of category
coordinates in semantic memory. However, in the absence of empirical evidence, this
remains an unresolved question for production models.
Recently, Hirschfeld and Zwitserlood (2012) employed the blocked cyclic
naming paradigm to test whether actions induced a semantic interference effect. The
paradigm involves small blocks of pictures (e.g., 4–6) presented repeatedly over
several cycles (e.g., 4–6). Related/homogeneous blocks usually comprise object
category exemplars (e.g., all animals) while unrelated/heterogeneous blocks comprise
pictures from different categories (e.g., animals, vehicles, furniture, fruit). Healthy
participants are typically slower to name objects in related compared to unrelated
blocks when they are repeated from the second cycle onward – a semantic
interference effect (Damian et al., 2001; Damian & Als, 2005; see Belke & Stielow,
2013 for review). It is generally accepted that the interference effect in blocked cyclic
naming originates during conceptual processing, with categorically related contexts
priming the activation levels of lexical candidates via feature sharing (see Belke,
2013; Oppenheim et al., 2010). The relative persistence of the effect has been
attributed to an incremental learning mechanism operating in the links between
conceptual and lexical representations (e.g., Damian & Als, 2005; Oppenheim et al.,
2010).1 In Hirschfeld and Zwitserlood’s (2012) experiment, related blocks comprising
pictures of different actions performed by the same body-part (hand, face and foot)
1 There is also debate about whether lexical selection occurs via competitive or non-competitive
mechanisms. The former mechanism assumes selection of the target utterance is made more difficult in
related contexts due to the priming of conceptual representations raising the lexical activation levels of
competitors (e.g., Belke, 2013; Damian et al., 2001). The latter assumes selection is accomplished
when a predetermined activation threshold or number of time steps is reached (e.g., Oppenheim et al.,
2010). The present study is primarily concerned with the conceptual representations engaged during
semantic interference effects in action naming rather than adjudicating between different lexical
selection mechanisms.
RUNNING HEAD: Interference from related actions 6
elicited a significant semantic interference effect. However, the stimuli included
depictions of both object-oriented actions (e.g., painting) and intransitive actions
accompanied by objects (e.g., singing via a microphone).
Evidence from lesion, neuroimaging and non-invasive brain stimulation
studies of blocked cyclic object naming has been incorporated in production models,
with key roles proposed for two left-hemisphere cortical regions: posterior middle and
superior temporal gyri (pMTG/STG) and inferior frontal gyrus (IFG) (e.g., Belke &
Stielow, 2013; Oppenheim et al., 2010; Schnur et al., 2009). There is relatively
consistent evidence that the left pMTG/STG area plays a role in mediating
conceptual-lexical processing. For example, the lesion-symptom mapping (LSM) and
perfusion neuroimaging studies of Harvey and Schnur (2015) and de Zubicaray et al.
(2014) show good agreement with clusters reported with peak maxima with Montreal
Neurological Institute (MNI) atlas coordinates of -52, -40, -5 and -46, -42, 2,
respectively for semantic interference. The non-invasive brain stimulation studies of
Pisoni et al. (2012), Krieger-Redwood and Jefferies (2014) and Meinzer et al. (2016)
likewise showed significant effects targeting similar MNI coordinates (-50, -46, 1 and
-54, -49, -2).
Reports of left IFG involvement in the block cyclic naming paradigm are
somewhat less consistent, and the proposed roles vary. Some neuroimaging studies
have observed differential activity for semantic interference (e.g., Schnur et al., 2009),
while others have not (e.g., de Zubicaray et al., 2014), and studies of aphasics with
left IFG lesions have produced different results for interference reflected in naming
latencies versus error rates (e.g., Biegler et al., 2008; Harvey & Schnur, 2015; Riès et
RUNNING HEAD: Interference from related actions 7
al., 2014; Schnur et al., 2009). Two anodal transcranial direct stimulation (aTDCS)
stimulation studies have shown short-lived facilitative effects of LIFG versus sham
stimulation on semantic interference (i.e., over the first four cycles only; Pisoni et al.,
2012; Meinzer et al., 2016), and a transcranial magnetic stimulation (TMS) study
reported an effect of LIFG stimulation in the first cycle only (Krieger-Redwood &
Jefferies, 2014). According to Schnur et al. (2009), left IFG biases interactions among
incompatible, non-target representations to help resolve lexical competition during
blocked cyclic naming. Belke and Stielow (2013) proposed a similar, top-down
account of LIFG involvement. Oppenheim et al. (2010) “tentatively” linked the left
IFG to a different mechanism that boosts all (i.e., target and non-target) lexical
activity until the difference between the most highly active candidate and the next
most active exceeds a threshold for selection.
It is worth noting interpretations of the neuropsychological evidence are
complicated by the potential involvement of two separate mechanisms in blocked
cyclic naming (see Belke & Stielow, 2013; Damian & Als., 2005; Krieger-Redwood
& Jefferies, 2014). Damian and Als (2005) were the first to propose a two-factor
account, noting naming latencies are occasionally faster in related blocks in the first
cycle, perhaps indicating a semantic priming mechanism, and the longer-lasting
semantic interference effect emerges only with repetition in subsequent cycles (see
also Belke & Stielow, 2013; Navarrete et al., 2014). Belke and Stielow (2013) have
also proposed the initial presentation cycle allows participants to establish or
memorise a task set (i.e., in terms of category membership), and then use this
information in subsequent cycles to bias selection. Yet, the majority of
neuropsychological studies have analysed data collapsed over all presentation cycles
RUNNING HEAD: Interference from related actions 8
(for review, see de Zubicaray et al., 2014). Hence, reports of left IFG and pMTG/STG
involvement from these studies potentially reflect contributions from more than one
mechanism. Recently, de Zubicaray et al. (2014) examined activity from the second
cycle onward with perfusion fMRI, observing significant differential signal changes
solely in the left pMTG/STG and the hippocampus. They interpreted the involvement
of the latter structure as reflecting the engagement of an incremental learning
mechanism (e.g., Damian & Als, 2005; Oppenheim et al., 2010; see Gluck, Meeter, &
Myers, 2003).
1.1 The present study
The purpose of the present study was to investigate whether a semantic
interference effect could be observed during bare naming of intransitive actions, and
to determine the nature and extent of the cognitive and neural mechanisms involved.
Our first behavioural experiment established a “pure” semantic interference effect
could be elicited with intransitive actions using the blocked cyclic naming paradigm.
In a second experiment using perfusion functional magnetic resonance imaging
(fMRI) with the same paradigm, we aimed to identify the neural mechanisms
involved.
Measuring cerebral perfusion changes during speech production with arterial
spin labeling (ASL) fMRI has several advantages over the more conventionally used
blood oxygen level dependent (BOLD) contrast mechanism (e.g., de Zubicaray et al.,
2014). For example, compared to BOLD signal, it is relatively insensitive to speech-
related motion-by-susceptibility artifacts in perisylvian cortical regions (e.g., Kemeny
RUNNING HEAD: Interference from related actions 9
et al., 2005; Detre et al., 2012; Liu & Brown, 2007). Perfusion fMRI also provides a
quantitative estimate of signal change that is more directly related to neural activity
(see Cavusoglu et al., 2012; Huppert et al., 2006). It additionally shows increased
sensitivity to group-level effects, due to the relatively smaller inter-individual
variability in perfusion compared to BOLD signal changes (Detre et al., 2012).
Within the production domain, neuroimaging, lesion and cortical stimulation
studies have provided converging evidence for left mid-posterior temporal cortex,
IPL, IFG and premotor cortex involvement in bare action naming (e.g., Breier &
Papanicolaou, 2008; Corina et al., 2005; Kemmerer et al., 2012; Liljeström et al.,
2008; Saccuman et al., 2006). Hence, all of these regions are plausible candidates for
mediating semantic interference effects during production of words denoting
intransitive actions. Modality specific activity in motor cortical areas during action
word comprehension is typically interpreted as supporting embodied/grounded
accounts of action meaning representation (for review, see Kemmerer, 2015; but see
de Zubicaray, Arciuli & McMahon, 2013 and Watson et al., 2013). However, it is
worth emphasizing that the aim of the current study is to determine whether a “pure”
semantic interference effect can be observed when naming intransitive actions in the
absence of object processing confounds, rather than to adjudicate between models of
embodied and amodal action concept representation. Indeed, as a number of authors
have pointed out, the mere detection of activity in motor cortical areas during action
word processing is insufficient to distinguish between these accounts (see de
Zubicaray, Arciuli, et al., 2013; Mahon & Caramazza, 2008; Watson et al., 2013).
RUNNING HEAD: Interference from related actions 10
Many neuroimaging studies of action comprehension have also shown activity
in left mid-to-posterior temporal cortex and inferior parietal lobe (IPL). These latter
regions are proposed to play roles of convergence zones whose activity reflects
processing of heteromodal, abstract representations of actions. For example,
Noppeney et al. (2005) reported that semantic decisions on action words increased
activation in the left anterior intraparietal sulcus (IPS) and mid-posterior temporal
cortex, and Van Dam et al. (2010) reported increased IPL activity for action verbs
associated with specific (e.g., to wipe) versus general (e.g., to clean) movements of
body parts. Thus, if these regions are involved in processing of action knowledge
during production, semantic interference in bare naming of intransitive actions should
be reflected in differential cerebral perfusion signal responses in posterior
temporoparietal regions. Finally, other regions implicated in more domain general
mechanisms during blocked cyclic naming, such as the left IFG and hippocampus, are
also expected to show differential activity (de Zubicaray et al., 2014; Harvey &
Schnur, 2015).
2. Experiment 1
2.1 Methods
2.1.1 Participants
Twenty-one healthy, native English-speaking adults participated (15 female,
mean age 20.29 years, range 17-29 years). All were undergraduate psychology
students who received partial course credit for participating in this experiment. All
had normal or corrected-to-normal vision. None reported any history of neurological
or psychiatric disorder, substance use, or hearing deficits. All provided informed
RUNNING HEAD: Interference from related actions 11
consent in accordance with the protocol approved by the Behavioural and Social
Sciences Ethical Review Committee of the University of Queensland.
2.1.2 Materials
A set of sixteen black-and-white line drawings served as targets, selected from
a range of action picture corpora and the internet (Druks & Masterson, 2000; Miozzo,
Fischer-Baum, & Postman, 2010; Szekely et al., 2004). Pictures comprised four
exemplars from each of four intransitive action contexts (face, arm, leg or whole-body
movements; Maouene et al., 2008) and were distributed orthogonally to create four
unrelated blocks (see Appendix). Transitivity was established using the online
Wordsmyth dictionary (Parks, Ray, & Bland, 1998).2 Blocks of four related (A) and
four unrelated (B) pictures were used to create counterbalanced lists of ABBA and
BAAB blocks in which trials were pseudo-randomly ordered such that no consecutive
items were identical or phonologically related (see Figure 1). Six presentation cycles
were created for each A and B block via Mix software (van Casteren & Davis, 2006),
with the requirement that consecutive trials never comprised the same picture or
phonological onsets.
2 Note that it is not possible to include a transitivity manipulation within the blocked cyclic naming
experiment without introducing a confound as the semantic interference effect is known to generalize
to novel items that share features/relations (e.g., Belke, Meyer, & Damian, 2005; see also Riès et al.,
2014). This would result in spread of semantic activation across related objects and actions/body parts
and to lexical representations, introducing a significant confound for interpreting a selective
interference effect for intransitive actions.
RUNNING HEAD: Interference from related actions 12
Figure 1. An example sequence of trials over two consecutive cycles in the blocked
cyclic naming paradigm employed in both experiments. The homogeneous context
shows trials from the leg action category; the heterogeneous context shows trials from
all categories.
2.1.3 Apparatus
Picture presentation and response recording were accomplished on a PC with a
15” display using the Cogent 2000 toolbox extension (v1.32;
www.vislab.ucl.ac.uk/cogent_2000.php) for MATLAB Software (The MathWorks
Inc., Massachusetts, USA). A Logitech Desktop Microphone with noise cancelling
technology was used to record responses on digital audio files. Naming latencies were
determined online with a voice-key implemented in the Cogent2000 toolbox. All
responses were verified off-line using Audacity software
(http://audacity.sourceforge.net).
2.1.4 Procedure
Participants completed a familiarization phase in which they named all 16
action pictures in random order, first with the correct gerundial form (e.g., laughing)
printed below and then without. The experimenter corrected participants if a mistake
was made. Two runs of 96 experimental items followed the familiarization phase,
with participants allowed a brief rest break in between. On each trial, a fixation cross
was presented for 500 ms, followed by the picture for 1500 ms, and a blank screen for
1000 ms. Participants were instructed to name each picture as quickly and as
accurately as possible using the gerundial form that names the action.
RUNNING HEAD: Interference from related actions 13
2.1.5 Analyses
We conducted repeated measures analyses of variance (ANOVAs) with
semantic context and presentation cycle as within participant variables, with
participants (F1) and items (F2) as random factors. Note that when item variability is
experimentally controlled by matching or by counterbalancing, which is the case in
the blocked cyclic naming paradigm, the traditional F1 is the most informative test
statistic (see Raaijmakers, Schrijnemakers, & Gremmen, 1999).
3. Results
Trials on which the voice key failed to detect a response or non-speech noises
triggered the voice key (N=4; 0.01%) were excluded from analyses. In addition,
correct trials with naming latencies deviating more than 2.5 standard deviations from
a participant’s mean response time (RT) within context (N=345; 8.55 %) were
considered outliers and excluded from analysis. Speech errors and dysfluencies were
rare (N=21; 0.52%). Due to the low rate, these errors were not subjected to analysis.
A total of 3662 trials were available for analysis. Figure 2 shows mean naming
latencies as a function of action meaning context (body-part related vs. unrelated) and
cycle. Mean naming latencies as a function of body-part relation are provided in
Supplementary Material Table S1.
The analysis revealed significant main effects of action meaning context [F1(1,
20) = 45.69, MSE = 1387.84, p < .001, ηρ² = .70; F2(1, 15) = 34.8, MSE = 1276.98, p
< .001, ηρ² = .70], such that response times were slower overall for the body-part
RUNNING HEAD: Interference from related actions 14
related (M = 589 ms) compared to unrelated context (M = 557 ms) and presentation
cycle [F1(5, 100) = 18.77, MSE = 884.382, p < .001, ηρ² = .48; F2(5, 75) = 17.28,
MSE = 772.45, p < .001, ηρ² = .54]. There was also a significant interaction [F1(5,
100) = 2.96, MSE = 1059.23, p = .016, ηρ² = .123; F2(5, 75) = 4.61, MSE = 562.9, p
= .001, ηρ² = .24]. As Figure 2 shows, naming latencies become slower from the
second cycle onward for the body-part related compared to unrelated sets, which is
the typical pattern seen for categorical object relations (see Belke & Stielow, 2013,
for a review). Following Belke and Stielow (2013) a second ANOVA was conducted
excluding data from the first cycle to determine if the interference effect was
cumulative over subsequent cycles (e.g., Oppenheim et al., 2010). This revealed a
significant main effect of action meaning context [F1(1, 20) = 74.50, MSE = 1020.86,
p < .001, ηρ² =.79; F2(1, 15) = 43.48, MSE = 1272.42, p < .001, ηρ² = .74] with body-
part related latencies again slower (584 ms vs. 546 ms) but not presentation cycle
[F1(4, 80) < 1, p = .50; F2(4, 60) < 1, p = .54]. The interaction was not significant
[F1(4, 80) < 1, p = .51; F2(4, 60) < 1, p = .46]. A paired t-test conducted on the means
(Mdiff = 0 ms, 95% CI = 29 ms) from the first presentation cycle was not significant [t1
< -1, p = .99; t1 < -1, p = .71].
RUNNING HEAD: Interference from related actions 15
Figure 2. Mean naming latencies as a function of context & cycle in Experiment 1.
Error bars are standard errors of the mean (SEM).
3.1 Discussion
We observed a significant interference effect in blocked cyclic action naming,
with slower naming latencies for body-part related actions. To our knowledge, this is
the first demonstration that related intransitive actions produce an interference effect
akin to categorical object relations in the blocked cyclic naming paradigm. However,
naming latencies did not differ in the first cycle or accumulate over cycles. It is worth
emphasising that these latter effects are also not consistently observed in the object
naming variant of the paradigm (e.g., Belke & Stielow, 2013).
4. Experiment 2
RUNNING HEAD: Interference from related actions 16
4.1 Methods
In Experiment 2, we sought to replicate the interference effect observed for
body-part related action contexts, and to investigate the neural mechanisms associated
with this effect. We therefore conducted a perfusion fMRI investigation. As
mentioned in the Introduction, our a priori hypotheses primarily targeted differential
perfusion responses in motor cortical and temporo-parietal regions that might reflect
grounded and/or heteromodal conceptual processing of actions, respectively.
4.1.1 Participants
Twenty-one healthy, native English-speaking adults participated (11 female,
mean age 23.33 years, range 19-30 years). All had normal or corrected-to-normal
vision. None reported any history of neurological or psychiatric disorder, substance
use, or hearing deficits. All were reimbursed AUD$30 for their participation. They
provided informed consent according to the protocol approved by the Medical
Research Ethics Committee of the University of Queensland. None participated in
Experiment 1.
4.1.2 Materials
Identical to Experiment 1.
4.1.3 Apparatus
Picture presentation and response recording were accomplished via a PC using
the Cogent 2000 toolbox extension for MATLAB as per Experiment 1. Pictures were
projected in black with a luminous white background onto a screen positioned at the
RUNNING HEAD: Interference from related actions 17
rear of the MRI system that participants viewed through a mirror mounted on the head
coil. The size of the pictures including background was approximately 10 cm wide by
10 cm high, and subtended approximately 10o of visual angle when each participant
was positioned for imaging. A 30 db attenuating headset was used to reduce gradient
noise. Naming responses were recorded on digital audio files using a custom
positioned fibre-optic dual-channel noise-cancelling microphone attached to the head
coil (FOMRI-III, Optoacoustics Ltd., Or-Yehuda, Israel;
http://www.optoacoustics.com). As per Experiment 1, naming latencies were
determined online with voice-key code implemented in the Cogent2000 toolbox, and
responses verified off-line using Audacity software (http://audacity.sourceforge.net).
4.1.4 Procedure
Participants completed a familiarization phase identical to Experiment 1. Two
runs of 96 experimental items followed the familiarization phase, with participants
allowed a brief rest break in between. On each trial, a fixation cross was presented for
500 ms, followed by the picture for 1000 ms, and a blank screen for 2000 ms. There
was a 4000 ms delay between blocks, during which a blank screen was shown. The
relatively longer inter-trial interval was employed in the fMRI experiment to assist in
resolving the perfusion response to each trial while the pause between blocks enabled
the perfusion response to return to baseline and avoided carry-over when switching
between homogeneous and heterogeneous contexts. Participants were instructed to
name each picture as quickly and as accurately as possible using the gerundial form.
4.1.5 Image Acquisition
RUNNING HEAD: Interference from related actions 18
Images were acquired using a 3 Tesla Siemens MAGNETOM Trio TIM
System (Siemens Medical Solutions, Erlangen, Germany) equipped with a 32-channel
receive-only phased-array head coil. Perfusion data were acquired using a quantitative
imaging of perfusion with a single subtraction, thin-slice TI1 periodic saturation
(Q2TIPS) with a proximal inversion with a control for off-resonance effects
(PICORE) labeling technique (Luh et al., 1999). The saturation slab was applied
inferior to the imaging slices, and was 20 mm thicker than the imaging slab, with a 10
mm margin at each edge, to ensure optimum inversion. In each of two consecutive
sessions, an initial M0 image followed by 152 interleaved control and label images
were acquired using a gradient-echo single shot echoplanar imaging (EPI) readout
with the following parameters: TI1 = 700 ms, TI2 = 1800 ms, TR/TE = 2500/11 ms,
matrix = 64×64, voxel in-plane resolution = 3×3 mm, flip angle = 90° and parallel
imaging (PI) reduction factor of 2 for optimal image quality (Ferré et al., 2012).
Volumes comprised 16 slices, 6 mm thick with a 1.5 mm gap, and were oriented to
ensure coverage of the whole cerebrum and most of the cerebellum. Prior to these
sessions, we elected to acquire a separate M0 image with a longer TR of 10000 ms to
maximize SNR for the equilibrium brain tissue magnetization used to normalize the
difference perfusion maps. The first five volumes of each 153 volume session
(consisting of the manufacturer’s M0 and two control and label images) were
discarded. Head movement was limited by foam padding within the head coil. A T1-
weighted structural image was acquired last using a magnetisation-prepared rapid
acquisition gradient-echo (MPRAGE) sequence (512 x 512 matrix, in plane resolution
.45 x .45 mm, 192 slices, slice thickness .9 mm, flip angle 7o, TI 1100ms, TR 2530
ms, TE 2.32ms).
RUNNING HEAD: Interference from related actions 19
4.1.6 Behavioural and Imaging Analyses
For the behavioural data, we conducted repeated measures ANOVAs with
semantic context (action related, unrelated) and presentation cycle (six cycles) as
within participant variables, all with participants (F1) and items (F2) as random
factors, in an identical manner to Experiment 1.
Image data preprocessing and analysis were conducted with the ASL toolbox
(ASLtbx; Wang, Z., et al., 2008) within statistical parametric mapping software
(SPM12; Wellcome Department of Imaging Neuroscience, Queen Square, London,
UK). Motion correction for the ASL image series was carried out using INRIalign
(Freire, Roche & Mangin, 2002), realigning subsequent images to the first image of
the first series. The realigned series were smoothed with a 6 mm FWHM isotropic
Gaussian kernel to reduce signal outliers by improving the spatial signal-to-noise ratio
(SNR) of both control and label images (Wang, Z., et al., 2008). The T1-weighted
image was segmented using the ‘New Segment’ procedure, and an intracranial mask
generated to exclude extracranial voxels for CBF calculation. Perfusion imaging time
series were then constructed for each participant by implementing a pairwise simple
subtraction between temporally adjacent label (tagged) and control acquisitions,
resulting in image volumes with an effective TR of 5 s (Liu & Wong, 2004). A mean
image was created from the perfusion time-series, and coregistered to the T1-
weighted structural image. The deformation fields produced by the ‘New Segment’
procedure for spatial normalisation to MNI atlas space were then applied to the
perfusion imaging time series, and volumes resliced to 2 mm3 voxels.
RUNNING HEAD: Interference from related actions 20
We conducted two-stage, mixed-effects model statistical analyses. At the
participant/fixed effects level, event types corresponding to the items in the two
experimental blocking contexts (action related and unrelated blocks) in each of the six
cycles were modeled as effects of interest with delta functions representing each
picture onset, and convolved with a synthetic haemodynamic response function
(HRF) for each session. First order time (i.e., linear) modulations for all event types
were included to accommodate between session variability. Error trials were modeled
separately as a regressor of no interest per session. Global perfusion signal
fluctuations were included per session as nuisance regressors to reduce between
session and between subject variability and enhance SNR (Wang, Z., 2012). In
addition, the segmented grey matter image from each participant was included as an
explicit mask. Temporal filtering was not employed due to its deleterious effects on
perfusion analyses (Wang, J., et al., 2005; Wang, Z., et al., 2008).
Linear contrasts were applied to each participant’s parameter estimates at the
fixed effects level. These contrasts were then smoothed with a 5 mm FWHM isotropic
Gaussian kernel to reduce between participant variability in brain structure and error
of voxel displacement during normalization (Wang, Z., et al., 2008) and entered in a
group level random effects repeated measures analysis of variance (ANOVAs) with
condition and cycle as within participant factors. Covariance components were
estimated using a restricted maximum likelihood (REML) procedure to correct for
non-sphericity (Friston et al., 2002). Our primary analyses involved planned contrasts
performed on correctly identified items according to blocking context and cycle
following the approach with the behavioural data. Specifically, we contrasted (1)
mean perfusion signal for related vs. unrelated blocks over all cycles; (2) mean
RUNNING HEAD: Interference from related actions 21
perfusion signal for related vs. unrelated blocks from cycle two onward (semantic
interference effect) and; (3) mean perfusion signal for related vs. unrelated blocks in
the first cycle only.
As we had a priori hypotheses concerning specific cortical regions associated
with various processing stages involved in action meaning representation and speech
production, we opted to first restrict voxel-wise analyses to a set of predefined regions
of interest (ROIs) via small volume corrections (SVC) within SPM12, thereby
controlling for multiple comparisons only in those voxels. The following ROIs were
selected from the Hammers et al. (2003) probabilistic atlas: left mid-temporal cortex
(mid-MTG/STG identified by the Indefrey & Levelt, 2004 meta-analysis associated
with lexical concept selection; see also Indefrey, 2011), LIFG (top down biasing or
booster mechanism; e.g., Belke & Stielow, 2013; Oppenheim et al., 2010; see Schnur
et al., 2009), and inferior parietal cortex (action word naming; e.g., Corina et al.,
2005; Liljeström et al., 2008; Saccuman et al., 2006). The left hippocampus
(incremental learning mechanism; de Zubicaray et al., 2014; see also Gluck et al.,
2003) and motor area ROIs (embodied action representations; Gallese & Lakoff,
2005; Kemmerer, 2015; Pulvermüller, 2005) were derived from the SPM Anatomy
toolbox (Eickhoff et al., 2006) based on cytoarchitectonic maximum probability maps
of Brodmann areas 6 and 4ap (Amunts, Schleicher, & Zilles, 2007).
We employed SVC as our hypotheses typically concerned a subset of voxels
within each ROI, rather than the mean activity across all voxels. However, by
estimating SVC thresholds from all voxels within the larger ROI, this approach
produces a more conservative threshold for controlling type 1 error. A height
RUNNING HEAD: Interference from related actions 22
threshold of p < .001 was adopted in conjunction with spatial cluster extent thresholds
of p < .05 (family-wise error [FWE] corrected) established independently for the
whole brain and each ROI.
5. Results
5.1 Behavioural data
Trials on which the voice key failed to detect a response or non-speech noises
triggered the voice key (N=54; 1.4%) were excluded from analyses. In addition,
correct trials with naming latencies deviating more than 2.5 SDs from a participant’s
mean RT within context (N=45; 1.1%) were considered outliers and excluded from
analysis. Speech errors and dysfluencies were rare (N=27; 0.67%). Due to the low
rate, these errors were not subjected to analysis. A total of 3852 trials were available
for analysis. Figure 3 shows mean naming latencies as a function of action meaning
context (body-part related vs. unrelated) and cycle. Mean naming latencies as a
function of body-part relation are provided in Supplementary Material Table S2.
The analysis revealed significant main effects of action meaning context [F1(1,
20) = 73.07, MSE = 530.25, p < .001, ηρ² = .79; F2(1, 15) = 56.46, MSE = 514.04, p
< .001, ηρ² = .79], such that response times were slower overall for the body-part
related (M = 750 ms) compared to unrelated context (M = 726 ms) and presentation
cycle [F1(5, 100) = 48.08, MSE = 570.89, p < .001, ηρ² = .71; F2(5, 75) = 33.94, MSE
= 655.3, p < .001, ηρ² = .69]. There was also a significant interaction [F1(5, 100) =
2.38, MSE = 1342.48, p = .044, ηρ² = .11; F2(5, 75) = 3.22, MSE = 371.45, p = .011,
ηρ² = .18]. As Figure 3 shows, naming latencies become slower from the second
RUNNING HEAD: Interference from related actions 23
cycle onward for the body-part related compared to unrelated sets. A second ANOVA
was conducted excluding data from the first cycle to determine if the interference
effect was cumulative over subsequent cycles (e.g., Oppenheim et al., 2010). This
revealed significant main effects of action meaning context [F1(1, 20) = 72.63, MSE =
567.57, p < .001, ηρ² =.78; F2(1, 15) = 46.6, MSE = 660.04, p < .001, ηρ² = .76] and
presentation cycle [F1(4, 80) = 3.50, MSE = 461.17, p = .011, ηρ² =.15; F2(4, 60) =
2.92, MSE = 473.71, p = .028, ηρ² = .16] with body-part related latencies again
slower (742 ms vs. 714 ms) and naming latencies becoming faster over cycles.
However, the interaction was not significant by participants [F1(4, 80) = 1.5, p = .22]
but was by items [F2(4, 60) = 2.74, MSE = 329.31, p = .037, ηρ² =.15]. A paired t-test
conducted on the means (Mdiff = 8.7 ms, 95% CI = 13.3 ms) from the first presentation
cycle was not significant [t1 = -1.36, p = .19; t2 = -1.27, p = .23].
5.2 Imaging data
5.2.1 A priori defined ROI analyses. Across all six cycles, comparisons of action
related vs. unrelated contexts revealed significant perfusion signal increases (i.e.,
action related > unrelated) and reductions (i.e., action related < unrelated) in multiple
ROIs. Increases were observed in left IFG, middle temporal cortex and hippocampus.
Perfusion signal decreases were observed in anterior middle and posterior temporal
cortex (see Table 1 and Figure 4).
RUNNING HEAD: Interference from related actions 24
Figure 3. Mean naming latencies as a function of context & cycle in Experiment 2.
Error bars are standard errors of the mean (SEMs).
For the first presentation cycle data, the left IFG and hippocampus ROIs
revealed significant perfusion signal increases (Figure 4). Significant perfusion signal
decreases were observed in the left anterior middle temporal cortex for the opposite
contrast (action related < unrelated). However, it should be noted that this contrast is
relatively underpowered as it involves (maximally) only 16 trials per condition per
participant. Comparisons involving data from the second cycle onward (i.e., the
semantic interference effect) again revealed significant perfusion signal increases
(i.e., action related > unrelated) in left IFG, hippocampus, and middle temporal
cortex. The opposite contrast revealed significant perfusion decreases in anterior
middle temporal and posterior temporal cortices and in the anterior intraparietal
sulcus (aIPS) (see Table 1 and Figure 5). No other ROIs showed significant
differential responses.
RUNNING HEAD: Interference from related actions 25
Table 1. Cerebral regions showing significant activity as a function of action context
(related vs. unrelated) and cycle in the fMRI experiment
Peak MNI
(x y z)
Z
score
Cluster Size
(Voxels)
Related > Unrelated Actions: All
Cycles
Left inferior frontal gyrusa,b
-42 42 -18 >8 8836
Right inferior frontal gyrusa 40 46 -2 7.08 1926
Left hippocampusb -14 -6 -26 6.71 283
Left middle temporal gyrusb -58 -28 -6 4.53 97
Related < Unrelated Actions: All
Cycles
Bilateral occipitotemporal cortexa 26 -68 -8 >8 18378
Left anterior middle temporal gyrusb -54 -4 -18 5.78 175
Left posterior middle and superior
temporal gyria, b
-44 -46 8 4.70 17
Related > Unrelated Actions: Cycle
1
Left inferior frontal gyrusa,b
-42 42 -18 6.11 2328
Left hippocampusb -14 -4 -28 3.91 28
Related < Unrelated Actions: Cycle
1
Bilateral Occipitotemporal cortexa 2 -98 20 7.81 9396
Left anterior middle temporal gyrusb -54 -4 -18 3.89 39
Related > Unrelated Actions: Cycles
2 to 6 (Semantic Interference)
Left inferior frontal gyrusa,b
-42 42 -18 >8 8437
Right inferior frontal gyrusa,b
40 46 -2 6.72 1427
Left hippocampusb -14 -6 -26 6.53 293
Left middle temporal gyrusb -58 -28 -6 4.17 40
Related < Unrelated Actions: Cycles
2 to 6 (Semantic Interference)
Bilateral Occipitotemporoparietal
cortexa
26 -68 -8 >8 8523
Left inferior parietal sulcusa, b
-32 -60 38 4.76 94
Left anterior middle temporal gyrusb -54 -4 -18 5.40 155
Left posterior middle and superior
temporal gyria, b
-44 -46 10 4.51 15
Height threshold p < .001 and p < .05 cluster FWE corrected aWhole brain corrected.
bROI corrected.
RUNNING HEAD: Interference from related actions 26
Figure 4. Cortical surface renderings showing (from left to right) perfusion increases
(i.e., action related > unrelated contexts) and decreases (i.e., action related < unrelated
contexts) in (top row) left lateral hemispheres over all cycles, and (bottom row)
during the first cycle. IFG = inferior frontal gyrus; aMTG = anterior middle temporal
gyrus; pMTG = anterior middle temporal gyrus. Responses are height thresholded at p
< .001 (uncorrected) and clusters > 50 voxels for visualization purposes.
5.2.2 Exploratory whole brain analyses. Across all six cycles, significant perfusion
increases and decreases were observed for the comparison of action related vs.
unrelated contexts, in large (i.e., spatially extensive) bilateral perisylvian cortical
networks. In addition to signal changes extending throughout the left hemisphere
ROIs noted above, signal increases were observed bilaterally extending throughout
the IFG (including pars orbitalis, pars triangularis and pars opercularis), rectal and
medial frontal gyri, and anterior hippocampi. Signal reductions were also observed in
both hemispheres, extending laterally and medially throughout occipital, posterior
temporal and parietal cortices, with a peak in the right fusiform gyrus (Table 1 and
Figure 4). For the first presentation cycle data, perfusion signal increases were
observed bilaterally in the IFG, in addition to portions of the medial and rectal frontal
RUNNING HEAD: Interference from related actions 27
gyri. The reverse contrast revealed signal decreases bilaterally throughout occipital,
posterior temporal and parietal cortices. For cycles two onward, signal increases were
observed bilaterally extending throughout the IFG (including pars orbitalis, pars
triangularis and pars opercularis), rectal and medial frontal gyri, and anterior
hippocampi. Signal reductions were also observed in both hemispheres, extending
laterally and medially throughout occipital and posterior temporal cortices and IPS,
with a peak in the right fusiform gyrus (Table 1 and Figure 5).
Figure 5. Cortical surface renderings showing (from left to right) perfusion increases
(i.e., action related < unrelated contexts) on lateral and medial views of the left
hemisphere (top row) and perfusion decreases (i.e., action related < unrelated
contexts) on lateral views of left and right hemispheres over cycles 2-6 (bottom row).
IFG = inferior frontal gyrus; aMTG = anterior middle temporal gyrus; pMTG =
anterior middle temporal gyrus; IPS = intraparietal sulcus. Responses are height
thresholded at p < .001 (uncorrected) and clusters > 50 voxels for visualization
purposes.
RUNNING HEAD: Interference from related actions 28
5.3 Discussion
We replicated the significant interference effect in naming latencies observed
in Experiment 1. The perfusion fMRI data over all cycles showed significant signal
changes in an extensive fronto-temporo-parietal cortical network for the contrast of
action related vs. unrelated contexts. Perfusion signal increases were observed
anteriorly in IFG and hippocampus, while decreases tended to be observed more
posteriorly in occipitotemporal and parietal (IPS) cortices in addition to anterior
temporal lobe. However, the motor area ROI did not show a significant context effect
for any of the contrasts.
6. General Discussion
In two experiments with a blocked cyclic paradigm, naming intransitive
actions in body-part related compared to unrelated contexts resulted in a significant
slowing of responses from the second cycle onward. This interference effect was
associated with significant perfusion signal increases and decreases in bilateral
cerebral networks encompassing predominantly frontal and medial temporal vs.
occipitotemporal and parietal cortices, respectively. No significant differences in
naming latencies according to context were observed in the first cycle in either
experiment. However, significant perfusion signal changes were observed in similar,
less extensive networks during the first cycle, and in more extensive networks when
all cycle data were combined. Below we discuss the implications of these findings for
models of spoken word production.
RUNNING HEAD: Interference from related actions 29
The novel finding of a reliable context effect across both experiments
indicates a “pure” semantic interference effect can be elicited during bare naming of
intransitive actions, i.e., the effect cannot be attributed to object feature confounds
that potentially occur with the use of transitive actions (e.g., Hirschfeld &
Zwitserlood, 2014). This semantic interference effect manifested from the second
cycle of naming onward, analogous to that observed for related object contexts (e.g.,
Belke & Stielow, 2013). Leaving aside the issue of the type of lexical selection
mechanism that might operate during blocked cyclic naming, production models
typically assume that semantic interference effects in naming have their origin in
conceptual representations or in the links between conceptual and lexical
representations (see Belke, 2013; Oppenheim et al., 2010).
Significant perfusion reductions were observed in a large bilateral network
encompassing lateral temporal cortex and intraparietal sulcus (IPS). These latter
regions have been proposed to play roles as convergence zones for processing of
heteromodal action meanings (e.g., Noppeney et al., 2005). A role for the anterior
temporal lobe in amodal conceptual processing across a variety of tasks is well-
established (see Binder & Desai, 2011). The peak in posterior MTG (-44, -46, 10)
accords well with those reported for object category coordinates in blocked cyclic
naming (e.g., -52, -40, -5 and -46, -42, 2; de Zubicaray et al., 2014; Harvey & Schnur,
2015), and suggests a processing mechanism in this region that is generic to naming
of both objects and actions. Reduced left MTG activity for semantic interference in
object naming has been observed across both PWI and blocked cyclic paradigms (e.g.,
de Zubicaray, Hansen, & McMahon, 2013; de Zubicaray et al., 2014; de Zubicaray &
McMahon, 2009; Piai et al., 2013, 2014). This relative decrease (related < unrelated)
RUNNING HEAD: Interference from related actions 30
in signal has been variously interpreted in terms of semantic priming (Piai et al.,
2014) or lateral inhibition between competing representations (de Zubicaray &
McMahon, 2009). The finding of increased perfusion signal accompanying the
semantic interference effect in another, more middle portion of the left MTG is
interesting as it suggests a processing distinction within this cortical structure, perhaps
reflecting a different mechanism for operations involving intransitive actions.
The differential perfusion signal in the anterior wall of the IPS (aIPS) is a
novel finding as this region has not been implicated in studies of the semantic
interference effect in blocked cyclic object naming.3 In his updated meta-analysis of
the neuroimaging data for spoken word production, Indefrey (2011) noted a
“probable, yet to date unclear role of the inferior parietal cortex”. This is perhaps
because the original Indefrey and Levelt (2004) meta-analysis collapsed data across
both object and action picture naming and word production studies. The present
findings clarify a role for the aIPS in conceptual-lexical processing of intransitive
actions during word production.
What type of conceptual-lexical processing might the aIPS engagement
reflect? Embodied accounts of action meaning representation propose that mirror
neurons in premotor cortex and IPL contribute to action understanding (Gallese &
Lakoff, 2005; Rizzolatti & Craighero, 2004). Consequently, the perfusion signal
3 As an anonymous reviewer noted, de Zubicaray et al. (2001) reported left IPL activity for a contrast
of semantically related distractors vs. a lexical control condition (a row of Xs) during object naming in
the PWI paradigm, i.e., a Stroop-like effect. Aside from the obvious differences in paradigm and
contrast employed, the current peak is 22 mm lateral and 26 mm posterior to that of the earlier PWI
result, and so in a macroanatomically (supramarginal gyrus vs. aIPS) and cytoarchitectonically (PFt vs.
hlP3) distinct region (see Caspers et al., 2006). Contrasts of related vs. unrelated distractors (i.e.,
semantic interference) during object naming in PWI typically do not elicit significant left IPL activity
(e.g., de Zubicaray & McMahon, 2009; de Zubicaray et al., 2013; Piai et al., 2013, 2014).
RUNNING HEAD: Interference from related actions 31
changes in the aIPS might be attributable to mirror neuron activity, and so reflect
modality-specific meaning activation. Although Rizzollati and Craighero (2004)
described intransitive actions as being capable of producing mirror system activation
in humans, they emphasized this activation was restricted to premotor cortex and did
not involve IPL, unlike transitive actions that activated both regions. Yet, premotor
cortex did not show significant perfusion signal changes associated with semantic
interference in the present study, reducing the likelihood that mirror neurons or motor
simulation were engaged (e.g., Gallese & Lakoff, 2005; Kemmerer, 2015;
Pulvermüller, 2005). Consequently, it seems unlikely that the perfusion changes
occurring in the aIPS could be attributed to a mirror neuron mechanism for action
meaning representation.
Neuroimaging and cortical stimulation studies have confirmed a role for the
IPL in representing action intentions, rather than actual movements. For example,
Desmurget et al., (2009) showed that electrical stimulation of left IPL regions in
awake surgical patients led to reports of an intention to move specific body parts
without movement, whereas stimulation of premotor cortex produced actual
movements. Using voxel-based lesion symptom mapping in stroke patients, Kalénine,
Shapiro and Buxbaum (2013) reported that processing action means (i.e., intention to
perform a particular movement) rather than outcomes (i.e., object related goals) relied
on the integrity of the left IPL. Thus, one possibility is that the aIPS activity during
semantic interference with intransitive actions reflects processing of a relatively
abstract nature, consistent with the proposals of prominent production models (e.g.,
Roloefs, 1992; Vigliocco et al., 2004), and spreading-activation accounts of
conceptual processing of actions (e.g., Mahon & Caramazza, 2008).
RUNNING HEAD: Interference from related actions 32
Indirect evidence for some of these regions’ involvement in processing of
action meaning during production comes from analyses of aphasic patients’ object
naming errors. For example, Schwartz et al. (2011) found posterior MTG and IPL
lesions were selectively associated with thematic errors during object naming (e.g.,
apple-worm, dog-bone), and proposed this reflected activation of action knowledge
linking the objects in an event context (e.g., eating). Neuroimaging studies of the PWI
paradigm have likewise shown differential activity in posterior MTG and IPL for
thematically related contexts with obvious action linkages (e.g., CHEESE-mouse; de
Zubicaray et al., 2013). However, thematic relations that do not emphasise actions in
obvious event contexts do not elicit significant interference or IPL activity in the
blocked cyclic naming paradigm (e.g., cowboy, wagon, rifle, buffalo; de Zubicaray et
al., 2014). This is consistent with Abdel Rahman and Melinger’s (2011) study that
showed interference in blocked cyclic naming only occurred for apparently unrelated
objects (e.g., stool, knife, bucket, and river) when the blocks were preceded by a
verbal cue describing an event context (e.g., fishing trip) and thus could be integrated
into a common theme.
Three other findings of potential interest deserve mention here. One
interesting result is the observation of significant perfusion increases in the IFG in all
analyses. The whole brain analyses indicated this was part of a larger cluster
extending into rectal and medial frontal gyrii. As reviewed in the Introduction to this
paper, the evidence for left IFG involvement in semantic interference in blocked
cyclic object naming has been less consistent than that for the pMTG. However,
actions typically take longer to name than objects (e.g., Szekely et al., 2005), and
RUNNING HEAD: Interference from related actions 33
might thus afford greater involvement of top down regulation when retrieving target
representations from among competing candidates. The present findings are
consistent with proposals that IFG is required for resolving lexical-semantic
competition, perhaps via a domain general mechanism that top-down biases selection
processes (e.g., Belke & Stielow, 2013; Harvey & Schnur, 2015; Schnur et al., 2009).
The relatively extensive IFG activity might also reflect greater difficulty in teasing
apart lexical activations due to the more abstract conceptual representations of
intransitive actions. Further, as IFG involvement was observed across both initial and
subsequent cycles, it clearly reflects a process (or processes) operating throughout
task performance, perhaps including a representation of the task itself (e.g., Belke &
Stielow, 2013). An interpretation in terms of a semantic priming mechanism operating
solely in initial cycles is also less likely due to the absence of a significant context
difference in naming latencies in the first cycle data across both experiments.
The second interesting finding is the observation of significant perfusion
increases in the left hippocampus. Hippocampal involvement has also been reported
for object category coordinates (de Zubicaray et al., 2014; Llorens et al., 2016).
Several authors have noted the need for production models to include a mechanism to
explain the persistence of semantic interference in paradigms such as blocked cyclic
naming, where the effect has been shown to survive intervening filler trials. One
proposed mechanism is incremental learning (e.g., Damian & Als, 2005; Oppenheim
et al., 2010). Neurophysiologically informed models have attributed a key role to the
hippocampus in incremental learning (see Gluck et al., 2003; Meeter et al., 2005).
Whether an incremental learning mechanism is necessary for the semantic
interference effect to emerge (e.g., Navarrete et al., 2014; Oppenheim et al., 2010), or
RUNNING HEAD: Interference from related actions 34
is simply engaged as a consequence of the blocked cyclic naming paradigm’s
repetition of items, remains to be demonstrated.
The third and final result of interest is the reduced perfusion signal responses
in the visual extrastriate cortices, also observed in PWI studies of semantic context
effects with object naming (e.g., de Zubicaray et al., 2013). This likely reflects
meaning dependent modulation of early perceptual processing (for a review, see
Collins & Olson, 2014). For example, sequential visual matching of objects is known
to be affected by semantic context (e.g., Gauthier et al., 2003), and category learning
has been shown to enhance visual perception of objects along dimensions relevant to
the learned categories (Folstein et al., 2015). Consequently, presenting intransitive
actions in categorically related versus unrelated contexts might enhance their
perceptual processing along coordinate dimensions (e.g., body part), resulting in
reduced perfusion signal.
6.1 Conclusions
Over two experiments, we observed a reliable semantic interference effect
during bare naming of intransitive actions in related versus unrelated contexts. This
novel finding may be interpreted as indicating conceptual representations of action
words are organized according to body part coordinate relations in semantic memory
(e.g., Maouene et al., 2008), and may thus inform future production models. The
interference effect was associated with perfusion increases and decreases in bilateral
cerebral networks encompassing predominantly frontal and medial temporal vs.
occipitotemporal cortices and the IPS, respectively. As the semantic interference
effect is generally assumed to have its origins in a conceptual preparation stage of
RUNNING HEAD: Interference from related actions 35
processing or in conceptual-to-lexical connections (e.g., Belke, 2013; Oppenheim et
al., 2010), the latter findings confirm a role for middle temporal cortex and IPS in the
conceptual-lexical processing of intransitive actions.
Acknowledgements
This research was supported by an Australian Research Council (ARC) Future
Fellowship FT0991634 (GZ). We are grateful to Michele Miozzo for providing some
of the action pictures.
References
Abdel Rahman, R., & Melinger, A. (2009). Semantic context effects in language
production: a swinging lexical network proposal and a review. Language and
Cognitive Processes, 24, 713-734.
Abdel Rahman, R., & Melinger, A. (2011). The dynamic microstructure of speech
production: semantic interference built on the fly. Journal of Experimental
Psychology: Learning Memory and Cognition, 37, 149-161.
Belke, E. (2013). Long-lasting inhibitory semantic context effects on object naming
are necessarily conceptually mediated: Implications for models of lexical-semantic
encoding. Journal of Memory and Language, 69, 228-256.
Belke, E., Meyer, A., & Damian, M. (2005). Refractory effects in picture naming as
assessed in a semantic blocking paradigm. Quarterly Journal of Experimental
Psychology A, 58, 667-692.
RUNNING HEAD: Interference from related actions 36
Belke, E., & Stielow, A. (2013). Cumulative and non-cumulative semantic
interference in object naming, Evidence from blocked and continuous manipulations
of semantic context. Quarterly Journal of Experimental Psychology, 66, 2135-2160.
Biegler, K. A., Crowther, J. E., & Martin, R. C. (2008). Consequences of an inhibition
deficit for word production and comprehension: Evidence from the semantic blocking
paradigm. Cognitive Neuropsychology, 25, 493-527.
Binder, J. R. & Desai, R. H. (2011). The neurobiology of semantic memory. Trends in
Cognitive Sciences, 15(11), 527-536.
Bird, H., Howard, D., & Franklin, S. (2000). Why is a verb like an inanimate object?
Grammatical category and semantic category deficits. Brain and Language, 72, 246–
309.
Breier, J. I., & Papanicolaou, A. C. (2008). Spatiotemporal patterns of brain activation
during an action naming task using magnetoencephalography. Journal of Clinical
Neurophysiology, 25(1), 7–12.
Caspers, S., Geyer, S., Schleicher, A., Mohlberg, H., Amunts, K., & Zilles, K. (2006).
The human inferior parietal lobule: cytoarchitectonic parcellation and interindividual
variability. Neuroimage, 33, 430-448.
RUNNING HEAD: Interference from related actions 37
Cavusoglu, M., Bartels, A., Yesilyurt, B., & Uludag, K. (2011). Retinotopic maps and
hemodynamic delays in the human visual cortex measured using arterial spin labeling.
Neuroimage, 59, 4044-4054.
Collins, J. A., & Olson, I. R. (2014). Knowledge is power: How conceptual
knowledge transforms visual cognition. Psychonomic Bulletin & Review, 21, 843-860.
Corina, D.P., Gibson, E.K., Martin, R., Poliakov, A., Brinkley, J., & Ojemann, G.A.
(2005). Dissociation of action and object naming: Evidence from cortical stimulation
mapping. Human Brain Mapping, 24, 1-10.
Damian, M. F., Vigliocco, G., & Levelt, W.J. M. (2001). Effects of semantic context
in the naming of pictures and words. Cognition, 81, B77–B86.
Damian, M. F., & Als, L. C. (2005). Long-lasting semantic context effects in the
spoken production of object names. Journal of Experimental Psychology: Learning
Memory and Cognition, 31:1372-1384.
Desmurget, M., Reilly, K. T., Richard, N., Szathmari, A., Mottolese, C., & Sirigu, A.
(2009). Movement intention after parietal cortex stimulation in humans. Science, 324,
811–813.
Detre, J. A., Rao, H., Wang, D. J., Chen, Y. F., & Wang, Z. (2012). Applications of
arterial spin labeled MRI in the brain. Journal of Magnetic Resonance Imaging, 35,
1026-1037.
RUNNING HEAD: Interference from related actions 38
de Zubicaray, G., Arciuli, J., & McMahon, K. (2013). Putting an "end" to the motor
cortex representations of action words. Journal of Cognitive Neuroscience, 25(11),
1957-1974.
de Zubicaray, G. I., Hansen, S., & McMahon, K. L. (2013). Differential processing of
thematic and categorical conceptual relations in spoken word production. Journal of
Experimental Psychology: General, 142, 131-142.
de Zubicaray, G., Johnson, K., Howard, D., & McMahon, K. (2014). A perfusion
fMRI investigation of thematic and categorical context effects in the spoken
production of object names. Cortex, 54, 135-149.
de Zubicaray, G.I., & McMahon, K.L. (2009). Auditory context effects in picture
naming investigated with event-related fMRI. Cognitive, Affective, & Behavioral
Neuroscience, 9, 260-269.
de Zubicaray, G.I., Wilson, S. J., McMahon, K. L., & Muthiah, S. (2001). The
semantic interference effect in the picture-word paradigm: an event-related fMRI
study employing overt responses. Human Brain Mapping, 14, 218-227.
Druks, J., & Masterson, J. (2000). An object and action naming battery. Hove:
Psychology Press.
RUNNING HEAD: Interference from related actions 39
Eickhoff, S. B., Stephan, K. E., Mohlberg, H., Grefkes, C., Fink, G. R., Amunts, K.,
& Zilles, K. (2005). A new SPM toolbox for combining probabilistic
cytoarchitectonic maps and functional imaging data. NeuroImage, 25, 1325-1335.
Ferré, J. C., Petr, J., Bannier, E., Barillot, C., & Gauvrit, J. Y. (2012). Improving
quality of arterial spin labeling MR imaging at 3 tesla with a 32-channel coil and
parallel imaging. Journal of Magnetic Resonance Imaging, 1239, 1233–1239.
Folstein, J., Palmeri, T.J., Van Gulick, A.E., Gauthier, I. (2015). Category learning
stretches neural representations in visual cortex. Current Directions in Psychology,
24(1), 17-23.
Freire, L., Roche, A., & Mangin, J. F. (2002). What is the best similarity measure for
motion correction in fMRI time series? IEEE Transactions on Medical Imaging, 215,
470-484.
Friston KJ, Glaser DE, Henson RNA, Kiebel S, Phillips C, and Ashburner J. Classical
and Bayesian inference in neuroimaging: applications. NeuroImage, 16, 484-512,
2002.
Gallese, V., & Lakoff, G. (2005). The brain’s concepts: The role of the sensory-motor
system in reason and language. Cognitive Neuropsychology, 22, 455-479.
RUNNING HEAD: Interference from related actions 40
Gauthier, I., James, T.W., Curby, K.M., Tarr, M.J. (2003). The influence of
conceptual knowledge on visual discrimination. Cognitive Neuropsychology, 20, 507-
523
Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in action.
Psychonomic Bulletin & Review, 9, 558-565.
Gluck, M. A., Meeter, M., & Myers, C. E. (2003). Computational models of the
hippocampal region: linking incremental learning and episodic memory. Trends in
Cognitive Sciences, 7, 269-76,.
Hammers, A., Allom, R., Koepp, M. J., Free, S. L., Myers, R., Lemieux, L., Mitchell,
T. N., Brooks, D. J., & Duncan, J. S. (2003). Three-dimensional maximum probability
atlas of the human brain with particular reference to the temporal lobe. Human Brain
Mapping, 19, 224-247.
Harvey, D., & Schnur, T.T. (2015). Distinct loci of lexical and semantic access
deficits in aphasia: Evidence from voxel-based lesion-symptom mapping and
diffusion tensor imaging. Cortex, 67, 37-58.
Hirschfeld, G., & Zwitserlood, P. (2012). Effector-specific motor activation
modulates verb production. Neuroscience Letters, 523, 1, 15-18.
Indefrey, P. (2011). The spatial and temporal signatures of word production
components: A critical update. Frontiers in Psychology 2:255.
RUNNING HEAD: Interference from related actions 41
Indefrey, P., & Levelt, W. J. M. (2004). The spatial and temporal signatures of word
production components. Cognition, 92, 101–144.
Kalénine, S., Shapiro, A.D., & Buxbaum, L.J. (2013). Dissociations of action means
and outcome processing in left-hemisphere stroke. Neuropsychologia, 51(7), 1224-
1233.
Kemeny, S., Ye, F. Q., Birn, R., & Braun, A. R. (2005). Comparison of continuous
overt speech fMRI using BOLD and arterial spin labeling. Human Brain Mapping,
24, 173-183.
Huppert, T. J., Hoge, R. D., Diamond, S. G., Franceschini, M. A., & Boas, D. A.
(2006). A temporal comparison of BOLD ASL and NIRS hemodynamic responses to
motor stimuli in adult humans. NeuroImage, 29, 368-382.
Kemmerer, D. (2015) Are the motor features of verb meanings represented in the
precentral motor cortices? Yes, but within the context of a flexible, multilevel
architecture for conceptual knowledge. Psychonomic Bulletin & Review. 22, 1068-
1075.
Kemmerer, D., Rudrauf, D., Manzel, K., & Tranel, D. (2012). Behavioral patterns and
lesion sites associated with impaired processing of lexical and conceptual knowledge
of actions. Cortex, 48, 826–848.
RUNNING HEAD: Interference from related actions 42
Krieger-Redwood, K., & Jefferies, E. (2014). TMS interferes with lexical-semantic
retrieval in left inferior frontal gyrus and posterior middle temporal gyrus: Evidence
from cyclical picture naming. Neuropsychologia, 64, 24-32.
Liljestr m, M., Tarkiainen, ., arviainen, T., ujala, ., Numminen, ., iltunen, .,
et al. (2008). Perceiving and naming actions and objects. Neuroimage, 41, 1132-1141.
Liu, T. T., & Brown, G. G. (2007). Measurement of cerebral perfusion with arterial
spin labeling: Part 1 Methods. Journal of the International Neuropsychological
Society, 13, 517-525.
Liu, T. T., & Wong, E. C. (2005). A signal processing model for arterial spin labeling
functional MRI. NeuroImage, 24, 207-215.
Llorens, A., Dubarry, A.-S., Trébuchon, A., Chauvel, P., Alario, F.-X., & Liégeois-
Chauvel, C. (2016). Contextual modulation of hippocampal activity during picture
naming. Brain and Language, 159, 92-101.
Luh, W. M., Wong, E. C., Bandettini, P. A., & Hyde, J. S. (1999). QUIPSS II with
thin-slice TI1 periodic saturation: a method for improving accuracy of quantitative
perfusion imaging using pulsed arterial spin labeling. Magnetic Resonance in
Medicine, 41, 1246–1254.
RUNNING HEAD: Interference from related actions 43
Mahon, B.Z. & Caramazza, A. (2008). A Critical Look at the Embodied Cognition
Hypothesis & a New Proposal for Grounding Conceptual Content. Journal of
Physiology - Paris, 102, 59-70.
Mahon, B. Z., Costa, A., Peterson, R., Vargas, K. A., & Caramazza, A. (2007).
Lexical selection is not by competition: A reinterpretation of semantic interference
and facilitation effects in the picture–word interference paradigm. Journal of
Experimental Psychology: Learning, Memory & Cogition., 33, 503–535.
Maouene, J. Hidaka, S. & Smith B. L. (2008). Body parts and early-learned verbs.
Cognitive Science, 32, 1200-1216.
Meeter, M., Myers, C. E., & Gluck, M. A. (2005). Integrating incremental learning
and episodic memory models of the hippocampal region. Psychological Review, 112,
560–585.
Meinzer, M., Yetim, O., McMahon, K., & de Zubicaray, G. (2016). Brain
mechanisms of semantic interference in spoken word production: An anodal
transcranial Direct Current Stimulation (atDCS) study. Brain & Language,157-158,
72-80.
Miozzo, M., Fischer-Baum, S., Postman, J. (2010). A selective deficit for inflection
production. Neuropsychologia, 48, 2427-2436.
RUNNING HEAD: Interference from related actions 44
Navarrete, E., Del Prado, P., Peressotti, F., & Mahon, B. Z. (2014). Lexical retrieval
is not by competition: evidence from the blocked naming paradigm. J. Mem. Lang.
76, 253-272.
Noppeney, U., Josephs, O., Kiebel, S., Friston, K. J., & Price, C. J. (2005). Action
selectivity in parietal and temporal cortex. Cognitive Brain Research, 25, 641–49
Oppenheim, G. M., Dell, G. S., & Schwartz, M. F. (2010). The dark side of
incremental learning: A model of cumulative semantic interference during lexical
access in speech production. Cognition, 114, 227-252.
Parks, R., Ray, J., & Bland, S. (1998). Wordsmyth English dictionary-thesaurus.
http://www.wordsmyth.net/.
Piai, V., Roelofs, A., Acheson, D. J., & Takashima, A. (2013). Attention for speaking:
domain-general control from anterior cingulate cortex in spoken word production.
Frontiers in Human Neuroscience, 7, 832.
Piai, V., Roelofs, A., Jensen, O., Schoffelen, J.M., & Bonnefond, M. (2014). Distinct
patterns of brain activity characterise lexical activation and competition in spoken
word production. PLoS ONE, 9(2), e88674.
Pisoni, A., Papagno, C., & Cattaneo, Z. (2012). Neural correlates of the semantic
interference effect: New evidence from transcranial direct current stimulation.
Neuroscience, 223, 56–67,.
RUNNING HEAD: Interference from related actions 45
Pulvermüller, F. (2005). Brain mechanisms linking language and action. Nature
Reviews Neuroscience, 6, 576-582.
Raaijmakers, J. G. W., Schrijnemakers, J. M. C., & Gremmen, F. (1999). How to deal
with “The language-as-fixed-effect fallacy”: common misconceptions and alternative
solutions. Journal of Memory and Language, 41, 416-426.
Riès, S. K., Greenhouse, I., Dronkers, N. F., Haaland, K. Y., & Knight, R. T. (2014).
Double dissociation of the roles of the left and right prefrontal cortices in anticipatory
regulation of action. Neuropsychologia, 63, 215-225.
Rizzolatti, G., & Craighero, L. (2004). The mirror-neuron system. Annual Review of
Neuroscience, 27, 169–192,.
Roelofs, A. (1993). Testing a non-decompositional theory of lemma retrieval in
speaking: Retrieval of verbs. Cognition, 47, 59-87.
Saccuman, M.C., Cappa, S.F., Bates, E.A., Arevalo, A., Della Rosa, P., Danna, M., &
Perani, D. (2006). The impact of semantic reference on word class: An fMRI study of
action and object naming. NeuroImage, 32, 1865-1878.
Schnur, T. T., Costa, A., & Caramazza, A. (2002). Verb production and the semantic
interference effect. Journal of Cognitive Science, 3, 1-26.
RUNNING HEAD: Interference from related actions 46
Schnur, T. T., Schwartz, M. F., Brecher, A., & Hodgson, C. (2006). Semantic
interference during blocked-cyclic naming: Evidence from aphasia. Journal of
Memory and Language, 54, 199-227.
Schnur, T. T., Schwartz, M. F., Kimberg, D. Y., Hirshorn, E., Coslett, H. B., &
Thompson-Schill, S. L. (2009). Localizing interference in blocked cyclic naming:
Convergent neuroimaging and neuropsychological evidence for the function of
Broca’s area. Proceedings of the National Academy of Sciences USA, 106, 322–327.
Schriefers, H., Teruel, E., & Meinshausen, R. M. (1998). Producing simple sentences:
Results from picture-word interference experiments. Journal of Memory and
Language, 39(4), 609–632.
Schwartz, M.F., Kimberg, D.Y., Walker, G.M., Brecher, A., Faseyitan, O., Dell, G.S.,
Mirman, D., & Coslett, H.B. (2011). A neuroanatomical dissociation for taxonomic
and thematic knowledge in the human brain. Proceedings of the National Academy of
Sciences, 108, 8520-8524.
Szekely, ., acobsen, T., D’ mico, S., Devescovi, ., ndonova, E., erron, D., et
al. (2004). A new on-line resource for psycholinguistic studies. Journal of Memory
and Language, 51, 247–250.
Szekely, ., D’Amico, S., Devescovi, A., Federmeier, K., Herron, D., Iyer, G.,
Jacobsen, T., Arévalo, A. L., Vargha, A., & Bates, E. (2005). Cortex, 41, 7-26.
RUNNING HEAD: Interference from related actions 47
van Casteren, M. & Davis, M.H. (2006). Mix, a program for pseudorandomization.
Behavior Research Methods, 38, 584-589.
van Dam, W. O., Rueschemeyer, S.-A., & Bekkering, H. (2010). How specifically are
action verbs represented in the neural motor system: an fMRI study. Neuroimage, 53,
1318-1325.
Vigliocco, G., Vinson, D.P, Lewis, W. & Garrett, M.F. (2004). Representing the
meanings of object and action words: The featural and unitary semantic space
hypothesis. Cognitive Psychology, 48, 422-488.
Vigliocco, G., Vinson, D.P., Siri, S. (2005). Semantic and grammatical class effects in
naming actions. Cognition 94, B91-100.
Wang, J. J., Alsop, D., Li, L., Listerud, J., Gonzalez-At, J., Schnall, M., & Detre, J. A.
(2002). Comparison of quantitative perfusion imaging using arterial spin labeling at
15 and 4 Tesla. Magnetic Resonance in Medicine, 48:242-54.
Wang, J. J., Wang, Z., Aguirre, G.K., & Detre, J. A. (2005). To smooth or not to
smooth? ROC analysis of perfusion fMRI data Magnetic Resonance Imaging, 23:75-
81,
Wang, Z., Aguirre, G. K., Rao, H., Wang, J., Fernández-Seara, M. A., Childress, A.
R., & Detre, J. A. (2008). Empirical optimization of ASL data analysis using an ASL
data processing toolbox: ASLtbx. Magnetic Resonance Imaging, 26, 261-269.
RUNNING HEAD: Interference from related actions 48
Watson, C., Cardillo, E., Ianni, G., & Chatterjee, A. (2013). Action concepts in the
brain: An activation-likelihood estimation meta-analysis. Journal of Cognitive
Neuroscience, 25, 1191–1205.
Appendix
U
nre
late
d b
lock
Related block
laughing yelling winking sneezing Face
pointing saluting clapping waving Arm/hand
running walking skating hopping Leg/Foot
swinging hiding resting climbing Whole body
Highlights
Naming of intransitive actions is slower in related versus unrelated contexts
Associated with perfusion fMRI signal changes in frontal and temporo-parietal regions
Frontal cortex activity may reflect domain general mechanisms for resolving interference
Temporo-parietal activity may reflect conceptual-lexical processing of actions