Semantic Context and Visual Feature Effects
in Object Naming: An fMRI Study using
Arterial Spin Labeling
Julia Hocking, Katie L. McMahon, and Greig I. de Zubicaray
Abstract
& Previous behavioral studies reported a robust effect of
increased naming latencies when objects to be named were
blocked within semantic category, compared to items blocked
between category. This semantic context effect has been attrib-
uted to various mechanisms including inhibition or excitation
of lexico-semantic representations and incremental learning
of associations between semantic features and names, and is
hypothesized to increase demands on verbal self-monitoring
during speech production. Objects within categories also share
many visual structural features, introducing a potential con-
found when interpreting the level at which the context effect
might occur. Consistent with previous findings, we report a
significant increase in response latencies when naming cat-
egorically related objects within blocks, an effect associated
with increased perfusion fMRI signal bilaterally in the hip-
pocampus and in the left middle to posterior superior tem-
poral cortex. No perfusion changes were observed in the
middle section of the left middle temporal cortex, a region
associated with retrieval of lexical–semantic information in
previous object naming studies. Although a manipulation of
visual feature similarity did not influence naming latencies, we
observed perfusion increases in the perirhinal cortex for
naming objects with similar visual features that interacted with
the semantic context in which objects were named. These
results provide support for the view that the semantic context
effect in object naming occurs due to an incremental learn-
ing mechanism, and involves increased demands on verbal
self-monitoring. &
INTRODUCTION
Nearly all modern theories of speech production make
the assumption that competition occurs naturally be-
tween a target word and its lexico-semantic neighbors
prior to its selection for articulation (see Levelt, Roelofs,
& Meyer, 1999). Ad hoc evidence for a competitive se-
lection process in normal speech is provided by analyses
of occasional errors showing semantic substitutions, in
which an intended word (e.g., warm) is spontaneously
replaced by a semantically related word (e.g., cold;
Harley & MacAndrew, 2001). Self-repairs of these errors,
or corrections of them without external prompting, im-
ply the existence of a monitoring process (or processes)
operating during speech production (Postma, 2000). Evi-
dence from spontaneous self-correction during speech
has shown that this monitoring process can occur at
different levels. For example, a word can be articulated
in full and then corrected, thus self-monitoring of overt
speech provides the speaker an opportunity to correct
any speech errors. Alternatively, only the first syllable
of a word may be spoken before being corrected. This
self-interruption at the onset of articulation suggests an
inner monitor that checks the correctness of the verbal
message at a phonological level (i.e., the output of pho-
nological encoding) prior to phonetic encoding and
articulation (Levelt et al., 1999; see also Indefrey & Levelt,
2004) (see Figure 1).
In a recent meta-analysis of the relevant neuroimag-
ing literature on word production, Indefrey and Levelt
(2004) identified the bilateral superior temporal gyri as
the possible neural correlates of verbal self-monitoring.
Although the meta-analysis was only able to directly
identify the correlates of the outer loop, an ‘‘economical
assumption’’ was made attributing the inner loop to
the same regions or to a subset of them (p. 125). Self-
monitoring has been predominantly investigated through
the experimental manipulation of auditory feedback
during overt speech. For example, using PET, McGuire,
Silbersweig, and Frith (1996) increased demands on
speech monitoring by giving normal, distorted, or alien
external feedback to subjects reading aloud nouns. This
enabled them to determine the brain regions involved
in a mismatch between subjects’ intended and perceived
verbal output. Increased activation was observed in
posterior and middle superior temporal cortices when
demands on self-monitoring were increased. Similarly,
Fu et al. (2006) manipulated auditory feedback in an
overt adjective naming task using fMRI. They reportedThe University of Queensland, Brisbane, Australia
D 2008 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 21:8, pp. 1571–1583

a network of regions engaged in self-monitoring, with
greater activation for correct recognition of self-generated
speech in the middle and superior temporal gyri.
Both the Fu et al. (2006) and McGuire et al. (1996)
studies reported bilateral superior temporal activation,
a result supported by additional studies that manipulat-
ed different types of verbal feedback (Allen et al., 2005;
Hirano et al., 1997). However, manipulations of phono-
logical or semantic characteristics of speech are reported
to elicit predominantly left-lateralized temporal activa-
tion (Binder et al., 1995; Demonet et al., 1992; Zatorre,
Evans, Meyer, & Gjedde, 1992). Are both the right and
left superior temporal regions responding to the process
of verbal self-monitoring? It is possible that previous
methods may have introduced an attentional confound
to their experiment due to the qualitative differences
between conditions in which the verbal signal is being
monitored. For example, when giving alien or distorted
verbal feedback to a subject, there is a mismatch be-
tween the actual spoken output and the expected ver-
bal feedback, that is, the subject expects to hear their
own voice but instead receives either a distorted signal
or someone else’s voice. The monitoring process be-
ing measured in these mismatched speech conditions
is thus not specific to verbal self-monitoring, being
confounded by increased attention to the unexpected
difference between self-generated speech and the ex-
perimentally modified verbal feedback signal (e.g., Fu
et al., 2006; McGuire et al., 1996). In contrast to the
bilateralactivationreportedinthesestudies,Alen
et al. (2005) found that verbal monitoring of degraded
speech increased activation solely in the left superior
temporal gyrus, a finding consistent with the role of
the left superior temporal gyrus in auditory verbal pro-
cessing (Scott, Blank, Rosen, & Wise, 2000). Importantly,
this effect was increased for self-generated speech but
attenuated during alien speech, suggesting that the
process of verbal self-monitoring (rather than monitor-
ing distorted or alien speech) more specifically engages
the left hemisphere. Taking all this evidence into ac-
count, we therefore predict that left (rather than bilat-
eral) superior temporal activation will be observed in a
task where self-monitoring of speech is increased while
attention and auditory feedback is held constant across
conditions.
According to the perceptual loop theory (Levelt,
1989), in addition to monitoring for speech errors, the
verbal self-monitoring system additionally monitors for
semantic correctness/appropriateness. In order to ma-
nipulate the demands on verbal self-monitoring, the pres-
ent study uses a semantic blocking paradigm to increase
the level of competition during speech production (Kroll
& Stewart, 1994). This paradigm entails a manipulation of
the context in which identical object pictures are named.
Behavioral studies have shown that naming latencies
increase when objects are grouped or blocked into se-
mantically related or homogeneous categories (e.g.,
chicken, whale, rabbit, giraffe, lamb) compared with
noncategorized or heterogeneous groups or blocks
(e.g., skirt, melon, tractor, cat, radish) (Ganushchak &
Schiller, 2008; Belke, Meyer, & Damian, 2005; Maess,
Friederici, Damian, Meyer, & Levelt, 2002; Vigliocco,
Lauer, Damian, & Levelt, 2002; Damian, Vigliocco, &
Levelt, 2001; Levelt et al., 1999; Kroll & Stewart, 1994).
When naming objects in semantically homogeneous
contexts, the self-monitoring system is presumed to be
engaged to a greater degree than in heterogeneous con-
texts (e.g., to check whether the correct alternative has
been chosen; see Ganushchak & Schiller, 2008; Maess
et al., 2002).
In the only neuroimaging study of the semantic block-
ing paradigm conducted to date, using MEG, Maess et al.
(2002) reported effects for naming in the homogeneous
context observed in the left temporal cortex. Firstly, they
reported an early effect in a time window of 150–
225 msec post picture onset that they attributed to the
mechanism responsible for the context effect. Secondly,
they reported a later effect also in the left temporal
cortex (450–475 msec post picture onset) that they
attributed to internal self-monitoring. Unfortunately, as
Maess et al. acknowledged, due to the limited source-
localization capabilities of MEG, their analysis did not
provide the real extent and shape of the activated
regions (p. 460). Consequently, they were not able to
identify which regions of the temporal cortex were re-
sponsible for the different effects.
At least three different mechanisms have been pro-
posed to account for the semantic context effect. Two
of these emphasize activation changes occurring at a
processing level involving retrieval of lexico-semantic
information. For example, increased naming latencies
could be due to suppression of activation of a target
word’s representations following naming of a recently
used, semantically related target (an inhibition account;
Vitkovitch, Rutter, & Read, 2001; McCarthy & Kartsounis,
2000). Alternatively, coactivation of lexical representa-
tions of semantically related items may occur, thus in-
creasing competition between a number of possible
Figure 1. The perceptual loop theory of speech monitoring
(Levelt, 1983, 1989; adapted from Slevc & Ferreira, 2006).
1572 Journal of Cognitive Neuroscience Volume 21, Number 8