doi:10.1093/brain/awv003
BRAIN 2015: 138; 1084–1096
| 1084
Verbal suppression and strategy use: a role for
the right lateral prefrontal cortex?
Gail A. Robinson,1,2 Lisa Cipolotti,2,3 David G. Walker,4 Vivien Biggs,4 Marco Bozzali5 and
Tim Shallice6,7
See Hornberger and Bertoux for a scientific commentary on this article (doi:10.1093/brain/awv027).
Verbal initiation, suppression and strategy generation/use are cognitive processes widely held to be supported by the frontal cortex.
The Hayling Test was designed to tap these cognitive processes within the same sentence completion task. There are few studies
specifically investigating the neural correlates of the Hayling Test but it has been primarily used to detect frontal lobe damage. This
study investigates the components of the Hayling Test in a large sample of patients with unselected focal frontal (n = 60) and
posterior (n = 30) lesions. Patients and controls (n = 40) matched for education, age and sex were administered the Hayling Test as
well as background cognitive tests. The standard Hayling Test clinical measures (initiation response time, suppression response
time, suppression errors and overall score), composite errors scores and strategy-based responses were calculated. Lesions were
analysed by classical frontal/posterior subdivisions as well as a finer-grained frontal localization method and a specific contrast
method that is somewhat analogous to voxel-based lesion mapping methods. Thus, patients with right lateral, left lateral and
superior medial lesions were compared to controls and patients with right lateral lesions were compared to all other patients. The
results show that all four standard Hayling Test clinical measures are sensitive to frontal lobe damage although only the suppression error and overall scores were specific to the frontal region. Although all frontal patients produced blatant suppression
errors, a specific right lateral frontal effect was revealed for producing errors that were subtly wrong. In addition, frontal patients
overall produced fewer correct responses indicative of developing an appropriate strategy but only the right lateral group showed a
significant deficit. This problem in strategy attainment and implementation could explain, at least in part, the suppression error
impairment. Contrary to previous studies there was no specific frontal effect for verbal initiation. Overall, our results support a role
for the right lateral frontal region in verbal suppression and, for the first time, in strategy generation/use.
1
2
3
4
5
6
7
Neuropsychology Research Unit, School of Psychology, The University of Queensland, Brisbane, Australia
Department of Neuropsychology, National Hospital for Neurology and Neurosurgery, Queen Square, London, UK
Dipartimento di Psicologia, University of Palermo, Italy
BrizBrain and Spine, The Wesley Hospital, Brisbane, Australia
Neuroimaging Laboratory, Santa Lucia Foundation, Rome, Italy
Institute of Cognitive Neuroscience, University College, London, UK
International School for Advanced Studies (SISSA), Trieste, Italy
Correspondence to: Gail Robinson,
School of Psychology,
The University of Queensland,
St Lucia,
Brisbane QLD 4072
Australia
E-mail: [email protected]
Keywords: Hayling Test; verbal suppression; inhibitory processes; strategy generation and use; frontal cortex, neuropsychology
Abbreviations: IFG = inferior frontal gyrus; LL/RL/SM = left lateral/right lateral/superior medial
Received July 1, 2014. Revised October 31, 2014. Accepted November 22, 2014. Advance Access publication February 9, 2015
ß The Author (2015). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved.
For Permissions, please email: [email protected]
Verbal suppression and strategy use
Introduction
Verbal initiation and suppression are executive functions
thought to be supported by the frontal cortex. These processes are typically measured with a variety of tasks such as
classic word fluency and Stroop tasks and inhibitory stopsignal tasks, which makes it difficult to draw conclusions
about their common and distinct neural correlates. The
Hayling Sentence Completion Test overcomes this limitation as it was specifically designed to assess verbal initiation
and suppression in the same task (Burgess and Shallice,
1996). The Hayling Test involves the oral presentation of
a sentence frame with the last word omitted. In Section 1,
individuals are required to complete the sentence with one
word that is sensible or meaningful (e.g. The captain stayed
with the sinking. . .ship). In Section 2 they must instead
complete the sentence with one word that is not semantically related to any word that is a natural completion
(e.g.. . .banana) or any other word in the sentence.
Generating a sensible completion requires verbal initiation
whereas generating an unconnected completion requires
suppression of ‘ship’ and the production of an unrelated
word such as ‘banana’. Performance on the Hayling Test
gives four scaled scores based on the response time to produce a meaningful word in Section 1 (initiation response
time) or a word that is not semantically related in Section 2
(suppression response time), the errors produced instead of
an unrelated completion in Section 2 (suppression errors)
and a combination of response times and errors in Sections
1 and 2 (overall score). A third key process integral to
successful performance on the Hayling Test is the ability
to generate and implement a strategy, which to date has
received little attention and is a major focus of this study.
The Hayling Test has been increasingly used as a measure of executive dysfunction for neurological patients with
a variety of pathologies (e.g. Parkinson’s disease: Bouquet
et al., 2003; Tourette’s: Channon et al., 2003; FTD/
Alzheimer’s disease: Hornberger et al., 2011; brain
tumour/stroke: Robinson et al., 2012; traumatic brain
injury: Senathi-Raja et al., 2010). In the original Hayling
study, Burgess and Shallice (1996) tested frontal (47 unilateral and 17 bilateral) and 27 posterior patients. In
Section 1, frontal patients produced longer response times
than both posterior patients and healthy controls. In
Section 2, frontal patients produced many more errors as
they fail to suppress generation of a natural completion or
a semantically connected word, compared to both posterior
patients and healthy controls (Burgess and Shallice, 1996;
Volle et al., 2012). Burgess and Shallice (1996) found that
healthy controls often rapidly learn to apply a strategy. The
two most frequently used strategies were choosing a visible
object in the room or linking the response to that in a
previous sentence frame (Bouquet et al., 2003; for a detailed discussion of strategy generation see Amati and
Shallice, 2007). By contrast, patients with frontal lesions
typically failed to use a strategy when producing an
BRAIN 2015: 138; 1084–1096
| 1085
unconnected word, even when response times were prolonged so allowing more ‘thinking’ time; instead they frequently produced semantically connected words. Moreover,
Burgess and Shallice (1996) inferred that the failure to
apply a strategy led to patients producing a greater
number of straight forward completion and semantically
connected errors.
Recently, there have been two small lesion studies investigating the neural correlates of verbal initiation and suppression. Volle et al. (2012) report 26 focal frontal patients,
subdivided into rostral (n = 8) and caudal (n = 18) lesions,
and non-frontal posterior patients (n = 19), compared to
healthy controls. They combined left/right lesions and
found that frontal patients were impaired compared to control subjects on each of the four scaled scores. Closer examination revealed that only the eight rostral frontal patients
were significantly worse on the initiation response time
(Section 1), but both rostral and caudal frontal patients
were worse on the suppression (Section 2) (response time
and errors). Volle and colleagues (2012) used two brain
lesion-deficit mapping methods, AnaCOM and voxel-based
lesion symptom mapping (VLSM), to compare patients to
controls (AnaCOM) and to other patients (VLSM). They
found specific associations between an initiation response
time deficit and right medial rostral frontal lesions
[Brodmann area (BA) 10], a suppression response time deficit
and right inferior frontal lesions (BA 45/47) and elevated suppression errors with orbitoventral frontal lesions (BA 11/32).
Volle and colleagues (2012) concluded that these three prefrontal regions may have distinct roles in the initiation and
suppression components of the Hayling Test.
In a different lesion study of frontal patients, Roca and
colleagues (2010) reported that a Hayling suppression error
deficit may be linked to the right anterior prefrontal region.
A short three-trial version of the suppression condition was
administered to 15 frontal patients. In this study, four of
the six lowest scoring patients had right lateral frontal lesions. However, some caution is warranted regarding the
critical lesions in these two studies as there were few patients with left inferior frontal lesions (Roca et al., 2010;
Volle et al., 2012) and left anterior lateral or right superior
lateral lesions (Volle et al., 2012). As Volle and colleagues
(2012) themselves note, a sufficient number of patients with
overlapping lesions is required for lesion-mapping methods
to be effective. Thus, the neural substrates for two of the
processes required for the Hayling Test (initiation and suppression) are not conclusively settled.
PET studies of healthy participants performing the
Hayling Test have implicated the left inferior frontal, left
middle/superior temporal and left operculum in response
initiation and the middle and inferior frontal gyri for response suppression (Nathaniel-James et al., 1997; Collette
et al., 2001). Allen and colleagues (2008) showed with
functional MRI that response suppression compared to initiation was associated with greater left middle and orbital
frontal and bilateral precuneus activation (see also de
Zubicaray et al., 2000), as well as greater prefrontal and
1086
| BRAIN 2015: 138; 1084–1096
temporal connectivity. These left prefrontal regions only
minimally converge with the few lesion studies that suggest
right prefrontal regions.
There are two key differences between imaging and lesion
studies that may explain why the findings diverge. One,
imaging studies require a large number of trials in contrast
to completing the standard Hayling Test that has only 15
trials in both conditions. Healthy individuals improve on
the suppression condition after the first few trials, presumably due to implementing a strategy, whereas frontal patients generally do not improve. Two healthy individuals
rarely produce suppression errors, unlike frontal patients.
They can reach a steady response state where they can
apply a strategy effectively. Hence, it may be the case
that imaging and lesion studies are testing different
processes.
In contrast to the focus on verbal initiation and suppression, strategy generation/use to facilitate production of a
word that is semantically unrelated to the sentence frame
has received very little attention. That is, a ‘new’ behaviour
given the context. How are new behaviours created or
chosen? Strategy generation/use impairments in frontal patients have been shown in at least two different paradigms,
including a spatial working memory task (Owen et al.,
1990). However, VLSM analysis of Tsuchida and Fellows
(2013) failed to find a significant frontal effect for the spatial search strategy score of this spatial working memory
task (Asselen et al., 2006). Reverberi and colleagues (2006)
reported impaired strategic processing in lateral frontal but
not medial frontal patients on a semantic word fluency
task. Thus, our knowledge of the role of individual frontal
regions in strategy generation and use is very limited from a
neuropsychological perspective. It is a key feature of the
current investigation, which also considers initiation and
suppression processes.
This study was designed to investigate performance on
the Hayling Test in patients with focal frontal and posterior
(non-frontal) lesions, and matched healthy control subjects,
in order to address the following aims: (i) to confirm the
sensitivity of the Hayling Test to frontal lobe damage by
comparing frontal patients to healthy controls and address
the contrasting evidence for frontal specificity of previous
lesion studies by comparing frontal to posterior patients;
(ii) to ascertain whether previous findings of a lack of lateralization (Burgess and Shallice, 1996) were corroborated,
we compared unilateral right and left frontal lesions; (iii) to
investigate whether the right lateral frontal findings for suppression errors of Roca et al. (2010) could be corroborated,
we used their methodology to analyse the effects of left
lateral, right lateral and superior medial lesions. We also
performed a novel comparison of patients with right frontal
lateral lesions to all other patients in a somewhat analogous procedure to the VLSM method, with the critical
group selected for a priori reasons (contrast analysis); (iv)
as the inferior frontal and orbital regions were specifically
thought critical for suppression errors by Volle et al.
(2012), we conducted a frontal gyri analysis for
G. A. Robinson et al.
suppression errors; and (v) for the first time, we use a
finer-grained lesion methodology to investigate the neural
correlates of strategy generation/use by analysing strategicbased responses on the suppression section of the Hayling
Test.
Materials and methods
Participants
Ninety-eight patients with focal frontal or posterior lesions
primarily due to brain tumour and stroke were recruited
from the National Hospital for Neurology and Neurosurgery
(NHNN) (London, UK) and BrizBrain and Spine (BBS) at the
Wesley Hospital (Brisbane, Australia). A description of patients’ lesion location, aetiology, lesion extent and chronicity
is available in Supplementary Table 1. Full details of inclusion
and exclusion criteria are detailed in Robinson et al. (2010,
2012). NHNN patients were recruited for several studies and
the same criteria were applied to BrizBrain and Spine recruitment. Eight patients were excluded, thus 90 patients with focal
lesions [60 frontal (43 NHNN: 16 BBS); 30 posterior: 15 left,
15 right (17 NHNN: 13 BBS)] were compared to 40 healthy
adult control subjects (33 UK; seven Australia) with no neurological or psychiatric history, matched as was possible for age,
gender and education.
Each participant provided informed written consent and approval for the study was given in the UK (National Hospital
for Neurology and Neurosurgery and the Institute of
Neurology Joint Research Ethics Committee, University
College London Hospitals NHS Trust Research and
Development Directorate) and Australia (The UnitingCare
Human Research Ethics Committee, The University of
Queensland Human Research Ethics Committee).
Baseline cognitive tests
The baseline cognitive tests comprised well-known clinical tests
with published standardized normative data collected from
large control samples. The National Adult Reading Test
(NART) was administered to estimate pre-morbid optimal
levels of functioning (Nelson and Willison, 1991). To assess
current level of general intellectual functioning, an untimed,
relatively culture-fair, non-verbal test of abstract reasoning
was used (Advanced Progressive Matrices; Raven, 1976). The
Graded Naming Test was used to assess nominal functions
(McKenna and Warrington, 1980).
Hayling Sentence Completion Test:
procedure and response scoring
criteria
The Hayling Sentence Completion Test was administered in
the published standard manner (Burgess and Shallice, 1997).
Participants were presented with 30 sentence frames with the
last word omitted and were required to generate a single word
to complete it under two conditions: Section 1 initiation – a
sensible and meaningful connected word (e.g. the captain
stayed with the sinking. . .ship; n = 15); or Section 2
Verbal suppression and strategy use
suppression – an unconnected word (e.g. for the previous sentence frame banana; n = 15). For both Sections 1 and 2, the
number of sentences completed and the total response time to
produce a word (i.e. end of sentence presentation to start of
word initiation) were recorded. The response time difference
between Sections 1 and 2 (i.e. Suppression – Initiation), in
addition to the four clinical Hayling scaled scores were calculated (initiation response time, suppression response time, suppression errors, overall scaled score). The scaled scores run
from 1 to 10, the points on the scale corresponding with percentiles: 1 = out of normal range or 5 1st percentile; 2 = 1st
percentile; 3 = 5th percentile; 4 = 10th percentile; 5 = 25th percentile; 6 = 50th percentile; 7 = 75th percentile; 8 = 90th percentile; 9 = 95th percentile; 10 = 99th percentile. In addition,
for the suppression condition, each response was coded into
one of eight possible categories for analysis of strategy-based
responses. Response scoring criteria for the suppression condition were based on those detailed by Burgess and Shallice
(1996), with some categories combined and additional categories added, as detailed in Table 1. We also calculated a
Global Error Score as detailed by Burgess and Shallice
(1996; Category A errors = 3 points, Category B errors = 1
point, maximum = 45; for results see Supplementary material).
The frequency of each response category was recorded for
each participant (full details in Supplementary material). In
addition, several composite measures were calculated including
an error ratio for Category A responses and correct Category
C responses with an obvious strategy (as detailed in Table 1).
These composite measures and the percentage of total
Category A, B and C responses are reported in the results
section.
Lesion analyses
A Neurologist (M.B.) who was blind to the history of each
patient and a neurosurgeon (D.W.) who was naive to the purpose of the lesion analysis reviewed the hard copies of MRI
scans (or CT where MRI was unavailable, n = 14). Brain MRI
BRAIN 2015: 138; 1084–1096
| 1087
was obtained on systems operated at 1.5 T, and included the
acquisition of an axial dual-echo, and an axial and coronal T1weighted scan. CT scans were all obtained using a spiral CT
system (SOMATOM PLUS 4, Siemens). Axial images were
collected with an effective slice thickness of 5 mm and pitch
of 1.5. Both MRI and CT data were used because our goal
was to enable the recruitment of a large number of patients.
The rigorous exclusion criteria and lesion assessment guidelines were based on detailed anatomical localization using
standard atlases (Duvernoy, 1991). Of note, all frontal lesions
were entirely within the frontal lobe except for two right frontal patients with vascular lesions that extended to the postcentral gyrus (analyses were conducted with and without these
two patients with the results being unchanged). The lesion localization method is described in detail in Robinson et al.
(2010, 2012). Briefly, each frontal patient was coded for the
presence of lesion and oedema in each hemisphere in the anterior and posterior portion of nine left and right frontal regions (18 areas in total). An area was only coded as damaged
if at least 25% was affected.
Three analyses (standard, lateralization, finer-grained) for
grouping frontal patients on the basis of their lesions were
carried out and were used for the behavioural analyses
(Supplementary material). In the standard analysis (frontal
versus posterior versus controls), all 18 regions were collapsed
together to establish a general frontal effect (frontal: n = 60;
posterior: n = 30; controls: n = 40). This analysis aimed to establish the sensitivity and specificity of the Hayling Sentence
Completion Test to the frontal lob. It was critical because only
if there was a significant frontal deficit compared to controls
on the Hayling Test overall and subscale scores, were subsequent analyses carried out. This is because the aim of these
further analyses was to be more specific anatomically on the
localization of key systems responsible for an already established frontal lesion effect.
Following the first standard analysis, the lateralization analysis was carried out by collapsing the nine left and right brain
regions and dividing the patients into left frontal (LF) and
Table 1 Description of the Hayling Test suppression section response type classification system
Response type Category (A, B, C) and Code
CATEGORY A: ERROR (Sensible Connected Responses)
1: Sensible responses
Error ratio: errors in the last 10/ errors in all 15 items
CATEGORY B: ERROR (Somewhat Connected Responses)
All Category B Errors
2: Semantic and/or opposite to response
3: Semantic to sentence
4: Semantic but bizarre
CATEGORY C: CORRECT (Nonsense Response)
All Correct Category C Responses
Category C responses with a strategy
5: Correct and visible
6: Correct and semantic to a previous response
7: Correct and both visible and semantic to previous response
8: Correct and no obvious strategy
Category description (and example)
Sensible sentence completion (The dough was put in the hot – oven)
Error ratio (number of errors last 10/ all 15)
Total number of somewhat connected errors (2-4)
Semantically connected to the expected sensible response (The dough was put
in the hot – sink / freezer)
Semantically connected to the subject of the sentence (The dough was put in the
hot – bread)
Response makes vague sense but in a bizarre/socially inappropriate way (The
whole town came to hear the mayor – cry)
Total number of correct responses (5-8)
Total number of correct responses with obvious strategy (5-7)
Item visible within the testing room (The dough was put in the hot – desk)
Semantically connected to a previous response (The dough was put in the hot –
orange; previous response: banana)
Meets criteria for both previous categories (The dough was put in the hot –
drawer; previous response: desk)
No obvious strategy used (The dough was put in the hot – train)
1088
| BRAIN 2015: 138; 1084–1096
right frontal (RF) groups, according to damaged hemisphere,
and contrasting them with controls (LF: n = 29, RF: n = 26,
Controls: n = 40). Note, five frontal patients with bilateral lesions were excluded. The lateralization analysis sought to investigate unilateral left and right frontal contributions to the
processes involved in the Hayling Test and to determine
whether collapsing over frontal regions (Burgess and Shallice,
1996; Volle et al., 2012) was justified. The results are reported
in the Supplementary material.
In parallel with the lateralization analysis, we carried out a
finer-grained analysis that was used by Roca and colleagues
(2010), allowing corroboration of their effects initiation. We
also note that this method is similar to other previous studies
(Stuss et al., 1998, 2005; Stuss and Alexander, 2007;
Robinson et al., 2012; Murphy et al., 2013), enabling comparison with behavioural analyses of frontal lobe dysfunction.
This analysis contrasted left lateral (LL) versus right lateral
(RL) versus superior medial (SM) versus controls (LL:
n = 15, RL: n = 11, inferior and middle frontal gyri; SM:
n = 14, superior frontal gyri; controls n = 40), to investigate
the localization of the processes underpinning the Hayling
Task within the frontal cortex. Primary lesion site was determined by a lesion falling entirely within one site or, when it
extended over two sites, the lesion was required to
G. A. Robinson et al.
affect 4 75% of the primary site and 5 30% of the secondary
site. Twenty frontal patients were excluded from this third
finer-grained analysis because the primary site of the lesion
fit more than one subgroup (n = 15) or because of the small
number of patients available for the inferior medial group
(n = 5; cingulate and orbital cortex) (Fig. 1).
A further novel contrast analysis was also used to address the right lateral frontal hypothesis. This contrast analysis
was based on the procedure of Alexander and colleagues
(2007) and examined the specificity of any right lateral effects,
contrasted with patients without right lateral damage (LL, SM,
Posterior). Thus patients with right lateral lesions (n = 11) were
contrasted with non-right lateral lesions (n = 59) in a somewhat analogous procedure to the VLSM method of Volle
et al. (2012). For the gyri analyses, frontal patients with any
part of their lesion in the anterior or posterior section of the
specific gyrus (inferior, middle or superior) were included in
the left versus right gyri analyses (for similar method, see
Robinson et al., 2010). This resulted in a higher number of
patients in each gyrus subgroup than in the finer-grained
subgroups as the criterion is inclusive rather than exclusive.
For example, the left and right inferior frontal gyrus subgroups
both comprised n = 17, compared to n = 15 and n = 11 for the
left lateral and right lateral subgroups, respectively.
Figure 1 Lesion location for patients with lesions predominantly affecting the right lateral, left lateral and superior medial
regions. The shaded areas represent the proportion of patients within each group who have lesions affecting at least 25% of the depicted region.
Verbal suppression and strategy use
Statistical analyses
Analysis of variance (ANOVA) was used for all analyses
(standard, lateralization, finer-grained, contrast). Age, premorbid intelligence and naming were included as covariates
because the right frontal group was significantly older than
the left frontal group, and the frontal patients scored marginally but significantly lower on the pre-morbid intelligence
measure than control subjects. In addition, patient groups
scored marginally but significantly lower than controls on
the naming test although, importantly, the performance of all
groups was well within the average range. Significant results
were followed by pairwise comparisons (standard and lateralization analyses: between patient groups and controls; finergrained analysis: each frontal subgroup compared to controls
using the Dunnett Test) (see previous studies for example of
method: Stuss et al., 1998; Troyer et al., 1998; Tucha et al.,
1999; Robinson et al., 2012). For the contrast analysis, the
right lateral and non-right lateral patient groups were directly
compared. Bonferroni’s correction was applied to the main
standard analyses (i.e. frontal versus posterior versus control
P 4 0.017) and to the lateralization and finer-grained analyses
too, but only if a frontal effect had not already been established in the standard analyses as the aim in these later analyses was to have more specific candidate regions for the locus
of an already established frontal effect (i.e. LF versus RF
versus Controls and LL/RL/SM versus Controls; both
P 4 0.017). Bonferroni’s correction was also applied to the
three gyri analyses (left versus right inferior/middle/superior
frontal; P 4 0.017). Following correction only significant results are reported. As the response time measures for the initiation and suppression Sections 1 and 2 were not normally
distributed, analyses were conducted on log transformations of
the data. If error variances were unequal, non-parametric statistics were applied (Turner et al., 2007; Robinson et al., 2012;
Volle et al., 2012). The chi-square test of independence was
used for the suppression error gyri analyses that specifically
compared the number of patients with lesions to specific gyri
who performed either within or below the age-related 5% cutoff reported in the Hayling Test manual by Burgess and
Shallice (1997).
The variety of aims required the use of multiple statistical
analyses on overlapping sets of data. Therefore in the interpretation of the results we use only the analyses relevant to
that aim, or if multiple aims are relevant to the interpretation
the standard analysis (for frontal-posterior comparisons) and
the contrast or, if not relevant, the finer-grained analysis.
Results
Descriptive characteristics and baseline cognitive tests
Descriptive characteristics and baseline cognitive test scores
are presented in Supplementary Table 2 for the patients
(posterior and frontal) and the healthy control subjects.
Control and patient groups were matched (i.e. P 4 0.05)
for sex, age and pre-morbid intelligence apart from the
older age of right frontal than left frontal patients
BRAIN 2015: 138; 1084–1096
| 1089
(Supplementary Table 4). In addition, the pre-morbid intelligence level (NART) of the Control group was marginally
higher than that of frontal patients (P 5 0.05) [standard
analysis: F(2,123) = 3.259, P 5 0.05]. Frontal and posterior
patients, and frontal subgroups, were equivalent in the time
since the lesion occurred.
For cognitive baseline measures the three basic groups in
the standard analysis did not differ in their performance on a
test of general intelligence [Advanced Progressive Matrices:
F(2,121) = 1.917, P 4 0.05]. There were no further frontal
effects [finer-grained analysis: F(3,70) = 0.573, P 4 0.05]
(Supplementary Table 2). The three basic groups differed
significantly on the naming measure [Graded Naming Test:
F(2,126) = 9.771, P 5 0.001] with both patient groups impaired, compared to Controls (frontal P 5 0.01; posterior
P 5 0.001)
[finer-grained
analysis:
F(3,75) = 3.293,
P 5 0.05, no specific LL, RL or SM effects]. Specifically,
85% of the frontal and 75% of posterior patients performed
above the 5% cut-off on the Graded Naming Test (95% of
controls performed above the cut-off on the naming test).
Overall performance on the Hayling
Test: frontal, posterior and control
groups
The Hayling Test overall score for the three basic groups
(frontal, posterior and control) differed significantly [nonparametric Kruskal-Wallis Test: H(2) = 29.359, P 5 0.001]
(Table 2). Posterior patients were unimpaired relative to
Controls (P 4 0.05), unlike the non-frontal posterior patients
in the Volle et al. (2012) study. By contrast, frontal patients
were impaired compared to both controls (P 5 0.001) and
posterior patients (P 5 0.05). Indeed 40% of frontal patients
obtained a score below the age-related 5% cut-off compared
with 13% of the posterior patients and 0% of the healthy controls, using the 5% cut-off scores outlined in Appendix 2 of the
Hayling Test manual (Burgess and Shallice, 1997) (see Table 3
and Supplementary material). In a similar fashion, the three
basic groups differed on the three Hayling subscale measures:
initiation response times [H(2) = 17.044, P 5 0.001; frontal
patients 5 Controls, P 5 0.001]; suppression response times
[H(2) = 18.723, P 5 0.001; frontal patients 5 Controls,
P 5 0.001, posterior patients 5 Controls, P 5 0.05]; and suppression errors [H(2) = 28.517, P 5 0.001; frontal
patients 5 Controls, P 5 0.001, posterior patients, P 5 0.01]
(Table 2 and Supplementary Tables 3–6). In addition, a higher
number of frontal patients performed below the age-related
5% cut-off, compared to both controls and posterior patients,
on each of the three Hayling subscales (Table 3 and
Supplementary material).
It is clear from the overall performance that the frontal
patients have substantial impairments on the Hayling Test.
We now turn to the suppression error subscale and report
the finer-grained and contrast results for the effects of left
lateral, right lateral and superior medial lesions (for the
Hayling initiation response time, suppression response
1090
| BRAIN 2015: 138; 1084–1096
G. A. Robinson et al.
time and all lateralization results see Supplementary material and Supplementary Table 4). It is notable, however, that
there is a significantly larger response time difference between the Hayling initiation and suppression sections for
frontal patients. This is most clear in the right lateral patients, compared to controls subjects, with superior medial
patients showing a milder effect [finer-grained analysis:
F(3,68) = 5.247, P 5 0.01; covariate age: P 5 0.01; RL
P 5 0.001; SM P 5 0.05] (Table 2 and Supplementary
Tables 3–6). A larger response time difference for frontal
patients corroborates the original Burgess and Shallice
(1996) finding, with further refinement in our study by
the dramatically larger right lateral deficit.
Suppression errors
As noted above, frontal patients were significantly impaired
on the suppression error subscale score compared to both
controls and posterior patients. If we examine the lesion location of frontal patients who performed within range or
below the age-related 5% cut-off, the difference between
left frontal and right frontal patients showed a trend
(P = 0.087) towards more right frontal patients below the
5% cut-off than within range and the reverse for the left
frontal patients (Fig. 2; see Fig. 3 for suppression error subscale scores by lesion extent). If we specifically examine the
left and right inferior, middle and superior frontal gyri separately for patients who perform within range or below the agerelated 5% cut-off, there is a significant effect only for lesions
to the inferior frontal gyrus (IFG) [2(2) = 6.195, P 5 0.05].
When left and right IFG lesions are directly compared, removing frontal patients without lesions to the IFG, this difference
remains significant [2(1) = 5.846, P 5 0.05]. Thus, for suppression errors, patients with right IFG lesions are more likely
to be 5 5% cut-off while patients with left IFG lesions are
more likely to perform within range.
Table 2 Hayling Sentence Completion Test
Healthy
controls
n = 40
Hayling Overall Scaled Scorea
Initiation response time Subscale Score
Number of sentences completed* (/15)
Suppression response
time Subscale Score a
Number of sentences completed (/15)
Response time difference
(Suppression-Initiation)
Total Response Timeb
Suppression errors Subscale Score
CATEGORY A: ERROR
(Sensible connected words)
All Category A
(%) 1. Sensible responses
Error ratio: last 10/ all 15
CATEGORY B: ERROR
(Somewhat connected words)
All Category B (%)
CATEGORY C: CORRECT
(Nonsense words)
All Category C responses
(5–8) (%)
Category C responses
with Strategy (5–7) (%)
Correct and no obvious strategy (%)
a
a
Posterior
patients
n = 30
6.05
(1.1)
5.73
(1.0)
14.98
(0.16)
5.63
(1.1)
14.95
(0.22)
25.9
(28.6)
5.17
(1.8)
5.33
(1.1)
14.97
(0.18)
4.77*
(1.6)
14.90
(0.40)
55.1
(64.5)
6.68
(1.3)
6.10
(1.8)
3.3
(5.9)
0.15
(0.35)
5.3
(7.7)
0.35
(0.45)
Frontal
patients
n = 60
3.68***,#
(2.2)
4.55***
(1.6)
14.82
(0.54)
4.07***
(2.0)
13.93*
(2.65)
74.0**
(105.0)
4.10***,#
(2.6)
Frontal subgroups
LL
RL
n = 15
n = 11
3.73***
(2.0)
4.40**
(1.8)
14.80
(0.41)
4.53*
(2.0)
13.80
(3.12)
76.5
(172.7)
4.53**
(2.3)
2.82***
(2.1)
5.00*
(1.1)
14.91
(0.30)
3.27***
(2.0)
14.18
(2.40)
113.2***
(102.2)
3.00***
(2.5)
SM
n = 14
4.57*
(2.0)
4.71**
(1.5)
14.93
(0.27)
4.57**
(1.7)
14.36
(1.45)
56.4*
(59.7)
5.00*
(2.6)
20.7***,#
(25.5)
0.45***
(0.41)
17.8**
(27.5)
0.46*
(0.46)
23.0***
(20.1)
0.49**
(0.45)
12.4*
(15.9)
0.36*
(0.41)
11.5
(12.3)
18.0
(15.8)
22.8**
(19.4)
21.3
(18.4)
36.3**
(23.3)
19.5
(17.3)
83.7
(17.8)
38.0
(29.1)
45.5
(26.5)
77.6
(22.2)
29.1
(24.8)
47.4
(21.5)
50.1***,#
(34.0)
22.3**
(26.3)
27.8**,#
(22.0)
51.5***
(32.1)
26.2
(29.9)
25.3
(17.9)
37.0***
(34.8)
10.9**
(14.4)
26.1
(25.9)
64.7*
(29.7)
29.5
(26.7)
35.3
(21.3)
Table shows Overall, Initiation Section 1 and Suppression Section 2: subscale scores, number of sentences completed, response time and response type (Category A, B and C) for
healthy controls and patients groups for the standard and finer-grained analyses (mean number or percentage, ratio of responses and standard deviations). *P 5 0.05, **P 5 0.01,
***P 5 0.001, compared to healthy controls.
#
Compared to the other patient group in the analysis.
a
Scaled Score is 1–10, 6 is average.
Reported in seconds but analysis based on Log transformation.
Verbal suppression and strategy use
BRAIN 2015: 138; 1084–1096
Category A connected words
Frontal patients produced a significantly higher percentage
of sensible completion words (Category A errors) than both
Table 3 Hayling overall and subscale scores
Section 1 Initiation RT: agerelated 5% cut-off
Section 2 Suppression RT: agerelated 5% cut-off
Section 2 Suppression errors:
age-related 5% cut-off
Overall Score: age-related 5%
cut-off
Frontal
patients
n = 60
Healthy
control
n = 40
Posterior
patients
n = 30
1
3
9***
0
5
24***
0
3
24***
0
4
24***
Number of healthy controls and patients (posterior and frontal) that performed below
the age-related 5% cut-off (Burgess and Shallice, 1997).
***P 5 0.005, compared to healthy controls and posterior patients.
RT = response time.
| 1091
Controls (P 5 0.001) and posterior patients (P 5 0.01)
[H(2) = 25.238, P 5 0.001], with no further specific frontal
effects [finer-grained analysis: H(3) = 17.751, P 5 0.001;
RL P 5 0.001, LL P 5 0.01, SM P 5 0.05] (Table 2 and
Supplementary Tables 3–6).
To investigate whether errors were produced throughout
the performance of the task that would be indicative of a
failure to produce and implement a useful strategy, we
analysed the error ratio measure that contrasted errors
for the last 10 items relative to errors on all 15 items.
The error ratio was designed to capture the ability to generate and use a strategy to produce appropriate words, as
Burgess and Shallice (1996) found that controls improve on
the suppression section after the first few trials. Frontal
patients produced a higher proportion of errors for the
last 10 items, relative to all 15 items, compared to
Controls [P 5 0.001; H(2) = 12.102, P 5 0.001]. All frontal subgroups were impaired [finer-grained analysis:
H(3) = 10.476, P 5 0.05; RL P 5 0.01, LL and SM both
P 5 0.05, respectively].
Figure 2 Hayling suppression errors. Age-related 5% cut-off and frontal lesion location. Bar charts show the number of patients within
range or below the 5% cut-off with a lesion absent or present. (A) All left frontal and right frontal regions; (B) left/right IFG; (C) left/right middle
frontal gyrus (MFG); and (D) left/right superior frontal gyrus (SFG). P-values for chi-square test of independence.
1092
| BRAIN 2015: 138; 1084–1096
Category B somewhat connected
words
Frontal patients produced a significantly higher percentage
of Category B suppression errors than Controls [P 5 0.01;
H(2) = 9.296, P 5 0.05]. Right lateral patients differed significantly from Controls (P 5 0.01) [finer-grained analysis:
H(3) = 13.540, P 5 0.01] (Table 2 and Supplementary
Tables 3–6). The right lateral group tended to give responses that were semantically connected to the correct
response or bizarre in meaning but all frontal subgroups
produced words that were semantic to the sentence
(Supplementary Table 5).
To address the critical role of the right lateral region,
right lateral and non-right lateral patients were directly compared. For the suppression error subscale
score the right lateral patients were impaired relative to
non-right lateral patients [F(1,83) = 6.839, P 5 0.05; covariates age and pre-morbid IQ P 5 0.001 and P 5 0.05,
respectively]. By contrast, for blatant Category A suppression errors or error ratio our right lateral and non-right
lateral patients did not differ [F(1,83) = 1.229, P 4 0.05;
age P 5 0.05 and F(1,83) = 0.257, P 4 0.05, respectively].
However, right lateral patients produced a significantly
higher percentage of Category B errors relative to
non-right lateral patients [F(1,83) = 8.369, P 5 0.01,
pre-morbid IQ P 5 0.05] (Table 2 and Supplementary
Tables 3–6).
Category C words and strategy use
Frontal patients (and all frontal subgroups) produced a significantly lower percentage of correct (unconnected) words,
compared to both controls and posterior patients
G. A. Robinson et al.
(P 5 0.001) [all Category C responses: H(2) = 30.168,
P 5 0.001;
finer-grained
analysis:
H(3) = 23.503,
P 5 0.001; RL and LL both P 5 0.001, SM P 5 0.05]. A
more specific pattern emerges when correct responses that
indicate the use of a strategy are examined. Only right lateral frontal patients produced a significantly smaller percentage of strategy responses, compared to controls
[finer-grained analysis: F(3,69) = 3.478, P 5 0.05; RL
P 5 0.01]. The strategy most often used by our controls
was to choose a visible object in the testing room, in line
with the findings of Burgess and Shallice (1996)
(Supplementary Table 3). However, if we examine correct
responses not comprising an obvious strategy, frontal patients produced a significantly smaller percentage of correct
but non-strategic responses than controls and posterior patients (both P 5 0.01) [correct and no obvious strategy:
F(2,117) = 8.891,
P 5 0.001;
finer-grained
analysis:
F(3,69) = 2.615, P = 0.058, not significant] (Table 2 and
Supplementary Tables 3–6).
When specifically investigating the right lateral frontal
hypothesis, right lateral patients produced a smaller percentage of correct responses than non-right lateral patients
[F(1,83) = 6.316, P 5 0.05; age P 5 0.001 and pre-morbid
IQ P 5 0.05] (Table 2 and Supplementary Tables 3–6).
However, right lateral and non-right lateral patients did
not differ in terms of strategic and non-strategic responses
[both P 4 0.05]. Of note, age was a significant covariate in
each strategic response analysis (Table 2 and
Supplementary Tables 3–6). Although the right frontal patients were older than the left frontal patients and therefore
age could have been a potential confound, the right lateral
deficit for suppression errors and also for the number of
correct responses compatible with a strategy remained significant after controlling for age.
Age and strategy use
Figure 3 Hayling suppression errors by lesion extent.
Hayling suppression errors by lesion extent for patients with right
inferior frontal (RIFG) damage or other (non-right IFG) frontal
damage (jittered equal values for ease of representation visually).
Given ‘Age’ was a significant covariate for a number of
strategy analyses, the relationship between ‘Correct responses with a Strategy’ and ‘Age’ was investigated with
correlation analysis. There was a significant negative correlation between age and strategic responses for the frontal
patients (r = 0.334, P 5 0.01) and healthy control subjects (r = 0.330, P 5 0.05). This suggests that with
increasing age, fewer strategic responses are produced, although this did not reach significance for posterior patients
(r = 0.270, P 4 0.05). There was also a significant negative correlation between strategic responses and the response time difference (i.e. suppression response time –
initiation response time) for frontal patients (r = 0.273,
P 5 0.01) and healthy controls (r = 0.497, P 5 0.01),
with a strong trend for posterior patients (r = 0.326,
P = 0.079). This is consistent with the notion of Burgess
and Shallice (1996) that the production of fewer strategybased responses was indicative of the patient not
implementing a strategy, which would also lead to longer
response times. By contrast, there was no relationship
Verbal suppression and strategy use
BRAIN 2015: 138; 1084–1096
between strategic responses and pre-morbid intelligence
(NART IQ) for any of the three basic groups (frontal:
r = 0.000; posterior: r = 0.337; control: r = 0.165, all
P 4 0.05) (Supplementary Fig. 1).
Discussion
To our knowledge, this is the first lesion study to investigate the distinct frontal neural correlates of strategy generation and use, as well as verbal initiation and suppression,
in the same task, the Hayling Task. Our large group of 60
frontal and 30 posterior patients is much larger than those
of the recent studies of Volle et al. (2012) (26 frontal, 19
posterior) and Roca et al. (2010) (15 frontal, no posterior)
that investigated initiation/suppression but comparable to
the size of the original Burgess and Shallice (1996) study.
Our main findings for verbal suppression and strategy
generation/use were of the presence of a right lateral
effect. Specifically, when required to produce unconnected
completions, patients with right lateral lesions produced a
higher number of semantically connected words (Category
B errors) and fewer correct responses that appear to be
derived from a strategy (Category C with strategy) than
the non-right lateral patients; in both cases these effects
were not shown by left lateral and superior medial cases.
Patients with right lateral lesions also had very much longer
suppression response times compared to non-right lateral
patients and controls. Left/right differences were revealed
by the frontal gyri examination with respect to suppression
| 1093
error deficits; these were more likely to be observed in patients with lesions to the right IFG than the left IFG. By
contrast, unlike the Volle et al. (2012) study, we did not
obtain a right medial rostral effect for verbal initiation.
In addition, our results confirmed the sensitivity of the
Hayling Test to frontal lobe lesions. Our sample of frontal
patients were impaired on all four clinical scaled scores
(overall score, initiation response times, suppression response times, suppression errors), compared to healthy controls. Also, we confirmed the specificity for two Hayling
measures (overall score, suppression errors) to frontal lesions as our frontal patients were impaired compared to
posterior patients. Our posterior patients had a largely
intact performance except for suppression response times;
they were similar in these respects to the posterior group in
the Volle et al. (2012) study.
A summary of our results in relation to our five aims is
presented in Table 4. We will now discuss our main findings in detail and in relation to the specific Hayling components (suppression, strategy generation and use,
initiation) and previous studies.
Verbal suppression
For the suppression error subscale score, only the frontal
patients were impaired when compared to both controls
and posterior patients. To illustrate the magnitude of the
effect, frontal patients were six times more likely than controls and four times more likely than posterior patients to
fail to suppress the completion word. The magnitude of
Table 4 Summary of main results for the five aims of the study
Aim
1. Hayling Test:
Sensitivity to frontal
lesions
Specificity to frontal
lesions
2. Hayling lateraliza
tion effects
3. Right lateral
frontal effect for
suppression
errors
4. Inferior frontal
gyrus effect for
suppression
5. Strategy analysis
for suppression
Section 2
Comparison groups
Significant results
Frontal versus Control
Frontal impaired compared to Control (RL, LL and SM all impaired): Overall scaled score,
initiation RT, suppression RT, suppression errors
Frontal impaired compared to posterior: Overall scaled score, suppression errors
Frontal versus Posterior
LF versus RF versus Control
RL versus non-RL patients
Right IFG versus Left IFG
A higher number of patients with right IFG lesions, compared to left IFG lesions, scored 5 5% agerelated cut-off for suppression errors (see Fig. 2).
Frontal versus Posterior versus
Control
LF versus RF versus Control
Frontal impaired compared to Control: Category C responses with strategy.
Frontal impaired compared to posterior and control: Correct and no obvious strategy.
RF impaired compared to Controls: Category C responses with strategy, Category C: correct
and visible, correct and no obvious strategy.
LF impaired compared to Controls: Category C: correct and no obvious strategy.
RL impaired compared to Controls: Category C responses with strategy, Category C: correct
and visible.
SM impaired compared to Controls: Category C: correct and both visible and semantic to
previous response.
RL/LL/SM versus Control
RT = response time.
RF impaired compared to LF: Category B: Semantic but bizarre
RF impaired compared to controls: RT difference, all Category B, Category C with strategy
LF impaired compared to RF and Controls: Initiation RT: no. sentences completed
RL impaired compared to non-RL: RT difference; suppression error subscale score; all
Category B errors; Category B: semantic to response, semantic but bizarre; all Category
C responses.
1094
| BRAIN 2015: 138; 1084–1096
total errors for all our groups was similar to that reported
by Burgess and Shallice (1996) (see Supplementary material
for global error scores). This adds validity to comparisons
between studies and weight to the reliability of the Hayling
Test across populations.
In their study Volle et al. (2012) reported that a right
orbitoventral region was the critical region for suppression
errors. This led us to select the right lateral as the potential
critical region a priori. Within our frontal patients, the right
lateral subgroup did indeed show the most severe deficits.
Right lateral patients failed to suppress the completion
word much more than healthy controls but to a similar
degree to left lateral and superior medial patients
(Category A error). However, there was a highly significant
right lateral effect for producing more semantically connected words (Category B error) instead of a ‘correct’ unrelated word. No such effect was found for the left lateral
or superior medial groups, who behaved in a similar fashion
to posterior patients. Thus right lateral patients specifically,
failed to realize that a word produced was subtly wrong.
The right lateral patients had lesions to the inferior and/
or middle frontal gyri. However, the strongest effects
involved the IFG; in that area we found a significant difference between right and left hemisphere patients. This
region is close to the right orbitoventral region identified
by Volle and colleagues (2012) in their AnaCOM and
VLSM analyses as critical for suppression errors. With
lesion studies it is difficult to be precise but there is a
high probability of lesion overlap between the two studies.
When the suppression error subscale score is considered,
our right lateral subgroup obtained a lower score of 3.00
(5th percentile) than the most impaired patients in the Volle
et al. (2012) study (rostral frontal = 4.25, i.e. 10–25th percentile); this suggests that the right IFG region damaged in
the right lateral group in our study is particularly critical.
By contrast, this region differs from the right anterior
region that Roca and colleagues (2010) linked to suppression errors; however, they used a much reduced version of
the task and had far fewer patients.
The critical right IFG region for suppression errors is
consistent with the broader lesion literature in which
damage to this area has been associated with inhibitory
failures on tasks such as the stop-signal one (Aron et al.,
2003). More recently, the association between the right IFG
and inhibitory processes has been further supported by ‘virtual lesion’ evidence with the use of transcranial magnetic
stimulation (Chambers et al., 2006; Verbruggen et al.,
2010) and in imaging and electrophysiological studies (for
detailed review see Aron, 2011; but see Hampshire et al.,
2010 for an alternative perspective).
Another factor that may be relevant, especially for the
semantically connected errors, is faulty monitoring. The
right lateral deficit, compared to non-right lateral patients
for semantically-related errors, suggests a special difficulty
with detection of less obvious errors. One possibility is that
it is more difficult to detect errors when responses only
subtly violate the task requirements in contrast to responses
G. A. Robinson et al.
that clearly violate the rules. Presumably an error monitoring threshold could be changed with damage to the critical
region. Indeed there has been much evidence of right lateral, particularly dorsolateral, frontal area involvement in
monitoring processes (Stuss et al., 2005; Fleck et al., 2006;
for review see Vallesi, 2012).
Strategies and the cause of the
suppression error impairments
With regard to the type of errors produced, all of our frontal subgroups (right lateral, left lateral, superior medial)
produced a significant number of blatant suppression
errors, that is giving a completion word and a higher proportion of these errors occurred in the last 10 sentences,
relative to errors on all sentences. The higher proportion
of errors late in the task is indicative of the patient not
developing a strategy for the task. Burgess and Shallice
(1996) indeed argued that the failure to attain an effective
strategy was a prime cause of high rates of completion and
semantically connected errors. This was based on a factor
analysis of the different types of responses made by their
patients. They found two main factors, one of which had a
strong and significant relationship to the absence or presence of a frontal lesion. This factor was strongly negatively
related to the rate of completion and semantically connected errors. It was also strongly positively related to the
number of correct responses that fitted a standard strategy,
but uncorrelated with the number of correct responses that
did not fit a standard strategy. Thus, Burgess and Shallice
(1996) argued that a failure in strategy attainment was central to high error rates because it would be much more
difficult for such a patient to inhibit a prepotent incorrect
response and generate a potentially correct response. If one
looks at the number of correct responses compatible with a
strategy, then the right lateral frontal subgroup showed a
significant specific deficit; they produced by far the lowest
percentage of such responses (11% of all responses by comparison with 28% for the other frontal subgroups, 29% for
posteriors and 38% for controls).
This suggests that the right lateral region has a key role
in generating or implementing an effective strategy. It might
be argued, however, that as the right lateral patients produced significantly more semantically connected words
(Category B errors), it is possible that they are actually
implementing a strategy, but an inappropriate one. There
is one reason to doubt that this account, an inhibitory failure account or a faulty monitoring one, could give a full
explanation of the right lateral problem. For verbal suppression, our right lateral patients had much longer
‘thinking’ times, reflected by the response time difference
between the suppression and initiation sections.
Moreover, the effect was very large. In the initiation section, right lateral patients were 60% slower than healthy
controls and insignificantly faster than non-right lateral patients. However, for the difference in response times
Verbal suppression and strategy use
between the suppression and initiation sections, the right
lateral patients were 4 400% slower than healthy controls
and nearly double non-right lateral patients. This is much
easier to explain from an inability to use an effective strategy and so being faced with the task of getting rid of an
inappropriate prepotent response than in either of the
other three accounts—faulty inhibition per se, or monitoring or use of a semantic strategy—all of which, if anything,
would give rise to faster responding.
Moreover, this localization is consistent with the only other
study to selectively localize a failure in strategy generation or
use within the frontal cortex. This is the study of Reverberi
et al. (2006) analysing strategy use in semantic fluency. It
differentiated between lateral and medial frontal lesions and
found the strategic failure only in the lateral group.
Why should the right lateral group have a particular difficulty in setting up an appropriate strategy? The strategies
used by subjects, such as looking round the room to find a
candidate object before hearing the sentence frame, are not
explicit or implicit in the task instructions. They require
what in the problem-solving literature has been called a
‘lateral transformation’ by Goel and Grafman (1995) or
‘restructuring’ of the problem by Ohlsson (1992). Lesions
to right ventrolateral prefrontal cortex appear to damage
the possibility of making a successful lateral transformation
in solving so-called Match Problems (Miller and Tippett,
1996; Goel and Vartanian, 2005).
Verbal initiation
For the Hayling Test Section 1 initiation response time
subscale, all three of our frontal subgroups were impaired
(Table 2). Thus, we failed to replicate the specific right
medial rostral deficit reported by Volle et al. (2012). We
did, however, identify a superior medial deficit that would
overlap with the right medial rostral area but there was
also an equivalent or stronger left lateral effect. One possible source for this difference is that there were few caudal
left frontal patients in the Volle et al. (2012) study (see
Fig1c, p.2430), in contrast to our 15 (caudal) left lateral
patients. Another explanation for our differing results may
lie in the severity of the initiation response time deficit as
Volle et al.’s rostral frontal group was slightly more impaired (subscale score = 3.62), relative to that of our frontal
groups (range: 4.38–5.00), possibly related to the very large
lesions of Volle’s rostral patients.
When considering verbal initiation processes, previous studies have focussed on initiation response times and not on a
failure to produce a word per se. In our study, of the nine
patients who failed to generate a completion word for all 15
sentences, seven had left frontal lesions and only one had a
right frontal lesion (see Supplementary Table 4 for significant
left frontal group deficit compared to right frontal patients
and controls). We note that this left frontal effect for verbal
initiation is consistent with the pattern documented in patients with specific verbal generation impairments like dynamic aphasia (for example of impaired Hayling Test
BRAIN 2015: 138; 1084–1096
| 1095
initiation in contrast to intact suppression with evidence of
strategy generation/use) (Robinson et al., 2005).
Frontal lateralization and the
Hayling Test
Finally, to summarize the findings above that address the
issue of frontal lateralization, our results suggest that when
left and right frontal patients are directly compared there is
a significant difference for two measures. The left frontal
patients completed fewer sentences than the right frontal
patients for the initiation Section 1 and, for the suppression
Section 2. Right frontal patients produced significantly
more semantically connected words that were bizarre
than left frontal patients (Supplementary Table 4). We
note that our specific right lateral effects, and in particular
the right IFG/left IFG contrast, suggest that the method
used in some previous studies collapsing over left and
right frontal patients are not justified as our results do
not corroborate a lack of lateralization.
General conclusion
In summary, our findings show that although all frontal
patients produce blatant suppression errors, there was a
specific right lateral frontal effect for producing semantically connected or ‘subtly’ incorrect responses. Moreover,
the longer ‘thinking’ times incurred by right lateral patients
reflect a lack of ability to generate and implement a strategy that was evident in patients producing fewer correct
responses comprising strategy-based responses. By contrast,
for verbal initiation, our findings did not support previous
studies proposing a right frontal role but rather there was a
suggestion for left lateral and superior medial roles.
Overall, the Hayling Test shows sensitivity to and specificity for frontal lobe damage and our findings suggest that
the right lateral frontal region is crucial for verbal response
suppression and strategy generation and use.
Funding
G.R. is supported by an Australian Research Council
Discovery
Early
Career
Researcher
Award
(DE120101119) and the Australian part of this study was
supported by the NEWRO Foundation (Australia).
Supplementary material
Supplementary material is available at Brain online.
References
Alexander MP, Stuss DT, Picton T, Shallice T, Gillingham S. Regional
frontal injuries cause distinct impairments in cognitive control.
Neurology 2007; 68: 1515–23.
1096
| BRAIN 2015: 138; 1084–1096
Allen P, Mechelli A, Stephan KE, Day F, Dalton J, Williams S, et al.
Fronto-temporal interactions during overt verbal initiation and suppression. J Cogn Neurosci 2008; 20: 1656–69.
Amati D, Shallice T. On the emergence of modern humans. Cognition
2007; 103: 358–85.
Aron AR. From reactive to proactive and selective control: developing
a richer model for stopping inappropriate responses. Biol Psychiatry
2011; 69: e55–68.
Aron AR, Fletcher PC, Bullmore ET, Sahakian BJ, Robbins TW. Stop
signal inhibition disrupted by damage to right inferior frontal gyrus
in humans. Nat Neurosci 2003; 6: 115–6.
Asselen M, Kessels RPC, Neggers SFW, Kappelle LJ, Frijns CJM,
Postma A. Brain areas involved in spatial working memory.
Neuropsychologia 2006; 44: 1185–94.
Bouquet CA, Bonnaud V, Gil R. Investigation of supervisory attentional system functions in patients with Parkinson’s Disease using
the hayling task. J Clin Exp Neuropsychol 2003; 25: 751–60.
Burgess PW, Shallice T. Response suppression, initiation and strategy
use following frontal lobe lesions. Neuropsychologia 1996; 34:
263–72.
Burgess PW, Shallice T. The Hayling and Brixton tests: test manual.
Bury St Edmunds, UK: Thames Valley Test Company; 1997.
Chambers CD, Bellgrove MA, Stokes MG, Henderson TR,
Garavan H, Robertson IH, Mattingley JB. Executive “brake failure”
following deactivation of human frontal lobe. J Cogn Neurosci
2006; 18: 444–55.
Channon S, Pratt P, Robertson MM. Executive function, memory, and
learning in Tourette’s Syndrome. Neuropsychology 2003; 17:
247–54.
Collette F, Van der Linden M, Delfiore G, Degueldre C, Luxen A,
Salmon E. The functional anatomy of inhibition processes investigated with the hayling task. Neuroimage 2001; 14: 258–67.
de Zubicaray GI, Zelaya FO, Andrew C, Williams SC, Bullmore ET.
Cerebral regions associated with verbal response initiation, suppression and strategy use. Neuropsychologia 2000; 38: 1292–304.
Duvernoy HM. The human brain: structure, three-dimensional sectional anatomy and MRI. New York: Springer-Verlag; 1991.
Fleck MS, Daselaar SM, Dobbins IG, Cabeza R. Role of prefrontal
and anterior cingulate regions in decision-making processes shared
by memory and nonmemory tasks. Cereb Cortex 2006; 16:
1623–30.
Goel V, Grafman J. Are the frontal lobes implicated in “planning”
functions? Interpreting data from the Tower of Hanoi.
Neuropsychologia 1995; 33: 623–42.
Goel V, Vartanian O. Dissociating the roles of right ventral lateral and
dorsal lateral prefrontal cortex in generation and maintenance of
hypotheses in set-shift problems. Cereb Cortex 2005; 15: 1170–77.
Hampshire A, Chamberlain SR, Monti MM, Duncan J, Owen AM.
The role of the right inferior frontal gyrus: inhibition and attentional
control. Neuroimage 2010; 50: 1313–19.
Hornberger M, Geng J, Hodges JR. Convergent grey and white matter
evidence of orbitofrontal cortex changes related to disinhibition in
behavioural variant frontotemporal dementia. Brain 2011; 134:
2502–12.
McKenna P, Warrington EK. Testing for nominal dysphasia. J Neurol
Neurosurg Psychiatry 1980; 43: 781–88.
Miller L, Tippett L. Effects of focal brain lesions on visual problemsolving. Neuropsychologia 1996; 34: 387–98.
Murphy P, Shallice T, Robinson G, MacPherson SE, Turner M,
Woollett K, et al. Impairments in proverb interpretation following
focal frontal lobe lesions. Neuropsychologia 2013; 51: 2075–86.
G. A. Robinson et al.
Nathaniel-James DA, Fletcher P, Frith CD. The functional anatomy of
verbal initiation and suppression using the Hayling Test.
Neuropsychologia 1997; 35: 559–66.
Nelson HE, Willison JR. The Revised National Adult Reading Test –
Test manual. Windsor, UK: NFER-Nelson; 1991.
Ohlsson S. Towards intelligent tutoring systems that teach knowledge
rather than skills: five research questions. N Dir Educ Technol 1992;
96: 71–96.
Owen AM, Downes JJ, Sahakian BJ, Polkey CE, Robbins TW.
Planning and spatial working memory following frontal lobe lesions
in man. Neuropsychologia 1990; 28: 1021–34.
Raven JC. Advanced Progressive Matrices, Set 1. Oxford, UK: Oxford
Psychologists Press; 1976.
Reverberi C, Laiacona M, Capitani E. Qualitative features of semantic
fluency performance in mesial and lateral frontal patients.
Neuropsychologia 2006; 44: 469–78.
Robinson G, Shallice T, Bozzali M, Cipolotti L. Conceptual proposition selection and the LIFG: neuropsychological evidence from a
focal frontal group. Neuropsychologia 2010; 48: 1652–63.
Robinson G, Shallice T, Bozzali M, Cipolotti L. The differing roles of
the frontal cortex in fluency tests. Brain 2012; 135: 2202–14.
Robinson G, Shallice T, Cipolotti L. A failure of high level verbal
response selection in progressive dynamic aphasia. Cognitive
Neuropsychol 2005; 22: 661–94.
Roca M, Parr A, Thompson R, Woolgar A, Torralva T, Antoun N,
et al. Executive function and fluid intelligence after frontal lobe lesions. Brain 2010; 133: 234–47.
Senathi-Raja D, Ponsford J, Schönberger M. Impact of age on longterm
cognitive
function
after
traumatic
brain
injury.
Neuropsychology 2010; 24: 336–44.
Stuss DT, Alexander MP. Is there a dysexecutive syndrome? Philos
Trans R Soc Lond B Biol Sci 2007; 362: 901–15.
Stuss DT, Alexander MP, Hamer L, Palumbo C, Dempster R,
Binns M, et al. The effects of focal anterior and posterior brain
lesions on verbal fluency. J Int Neuropsychol Soc 1998; 4: 265–78.
Stuss DT, Alexander MP, Shallice T, Picton TW, Binns MA,
Macdonald R, et al. Multiple frontal systems controlling response
speed. Neuropsychologia 2005; 43: 396–417.
Troyer AK, Moscovitch M, Winocur G, Alexander MP, Stuss D.
Clustering and switching on verbal fluency: the effects of focal frontal- and temporal-lobe lesions. Neuropsychologia 1998; 36:
499–504.
Tsuchida A, Fellows LK. Are core component processes of executive
function dissociable within the frontal lobes? Evidence from humans
with focal prefrontal damage. Cortex 2013; 49: 1790–800.
Tucha OW, Smely CW, Lange KW. Verbal and figural fluency in patients with mass lesions of the left or right frontal lobes. J Clin Exp
Neuropsychol 1999; 21: 229–36.
Turner MS, Cipolotti L, Yousry T, Shallice T. Qualitatively different
memory
impairments
across
frontal
lobe
subgroups.
Neuropsychologia 2007; 45: 1540–52.
Vallesi A. Organisation of executive functions: Hemispheric asymmetries. J Cogn Psychol 2012; 24: 367–86.
Verbruggen F, Aron AR, Stevens MA, Chambers CD. Theta burst
stimulation dissociates attention and action updating in human
inferior frontal cortex. Proc Natl Acad Sci USA 2010; 107:
13966–71.
Volle E, Costello AL, Coates LM, McGuire C, Towgood K, Gilbert S,
et al. Dissociation between verbal response initiation and suppression after prefrontal lesions. Cereb Cortex 2012; 22: 2428–40.