The Different Neural Correlates of Action
and Functional Knowledge in Semantic
Memory: An fMRI Study
Nicola Canessa
1,2,3
, Francesca Borgo
4
, Stefano F. Cappa
1,2,5
,
Daniela Perani
2,5
, Andrea Falini
5
, Giovanni Buccino
3
, Marco
Tettamanti
2,5
and Tim Shallice
4,6
1
CRESA, San Raffaele Scientific Institute, 20031, Milan, Italy,
2
Center for Cognitive Neuroscience, San Raffaele Scientific
Institute, 20132, Milan, Italy,
3
Neurosciences Department,
University of Parma, 43100, Parma, Italy,
4
Cognitive
Neuroscience Sector, SISSA, 34014, Trieste, Italy,
5
CERMAC,
Vita-Salute San Raffaele University, 20132, Milan, Italy and
6
Institute of Cognitive Neuroscience, University College,
London, WC1E 6BT, UK
Previous reports suggest that the internal organization of semantic
memory is in terms of different ‘‘types of knowledge,’’ including
‘‘sensory’’ (information about perceptual features), ‘‘action’’ (motor-
based knowledge of object utilization), and ‘‘functional’’ (abstract
properties, as function and context of use). Consistent with this
view, a specific loss of action knowledge, with preserved func-
tional knowledge, has been recently observed in patients with left
frontoparietal lesions. The opposite pattern (impaired functional
knowledge with preserved action knowledge) was reported in
association with anterior inferotemporal lesions. In the present
study, the cerebral representation of action and functional knowl-
edge was investigated using event-related analysis of functional
magnetic resonance imaging data. Fifteen subjects were presented
with pictures showing pairs of manipulable objects and asked
whether the objects within each pair were used with the same
manipulation pattern (‘‘action knowledge’’ condition) or in the same
context (‘‘functional knowledge’’ condition). Direct comparisons
showed action knowledge, relative to functional knowledge, to
activate a left frontoparietal network, comprising the intraparietal
sulcus, the inferior parietal lobule, and the dorsal premotor cortex.
The reverse comparison yielded activations in the retrosplenial and
the lateral anterior inferotemporal cortex. These results confirm
and extend previous neuropsychological data and support the
hypothesis of the existence of different types of information
processing in the internal organization of semantic memory.
Keywords: anterior inferotemporal cortex, conceptual knowledge,
functional semantic features, inferior parietal lobule, manipulative
semantic features
Introduction
How is semantic knowledge organized and how it is repre-
sented in the brain are central questions of cognitive neurosci-
ence. The investigation of patients who, after a brain lesion, are
affected by disorders of semantic memory, which selectively
affect knowledge of living or nonliving items, has played
a central role in this area of research. Although the existence
of category-specific semantic impairment is relatively uncon-
troversial, the interpretation of the findings derived from the
study of these patients remains a matter of discussion. A number
of alternative hypotheses have been proposed, and the debate
has been extensively reviewed in several recent publications
(see, e.g., Laiacona et al. 2003; Thompson-Schill et al. 2003;
Gainotti 2004, 2005). A basic distinction is between theories
that propose that such category-specific impairments reflect
the taxonomical, domain-specific organization of semantic
memory (Caramazza and Shelton 1998) and ‘‘reductionist’’
interpretations, which interpret the semantic category effect
as a by-product of other noncategorical aspects of conceptual
representations. One of the most well-known approaches of the
latter type is based on the idea that the internal organization of
semantic memory is in terms of different ‘‘types of knowledge’’
(Warrington and Shallice 1984). These include ‘‘perceptual’’
(comprising information about perceptual features) and ‘‘func-
tional’’ (including abstract and propositional properties, such as
function, location, and context of use) knowledge. According to
the original version of this model, category-specific deficits
reflect the differential weighting of these types of knowledge in
the semantic representations. Semantic knowledge of animals
and other living things, which share a number of characteristics,
relies more strongly on perceptual information, whereas
knowledge of artifacts, which may share the same function
despite remarkably different appearance, is primarily based on
functional properties (Warrington and Shallice 1984). This
‘‘perceptual--functional’’ model was subsequently extended to
account for a specific deficit at identifying smaller and manip-
ulable items, such as kitchen tools, relative to larger non-
manipulable man-made objects, such as vehicles. The revised
model underlined the importance of the dominant channel of
experience (perceptual or motor) associated with the acquisi-
tion, storage, and retrieval of the semantic representation of a
given object (Warrington and McCarthy 1987).
As to the neural correlates of semantic knowledge, ‘‘feature-
based models’’ are usually grounded in the proposal that object
concepts may be represented in the brain as distributed
networks of activity in the areas involved in the processing of
perceptual or functional knowledge. Indeed, an extensive
literature suggests that information about different object
features (i.e., different ‘‘types of knowledge’’) may be stored in
distinct regions of the cortex. In particular, different activations
for the retrieval of functional versus perceptual knowledge
(Cappa et al. 1998; Mummery et al. 1998; Thompson-Schill et al.
1999) and perceptual versus manipulative (Phillips et al. 2002)
and for specific attributes (color, form, motion) (Chao et al.
1999) have been reported (see Martin and Chao 2001 and
Thompson-Schill 2003, for a review).
Although the notion of perceptual features can be easily
defined (see, e.g., Vinson et al. 2003), the definition of functional
features has been used in a more loose way, to include widely
different classes of information. Indeed, it is worth noting that
the term functional does not refer exclusively to the knowledge
of object’s function but rather to those abstract propositional
properties of concepts that do not belong to the perceptual or
the motor domains (Martin and Chao 2001), for instance, the
‘‘context of use.’’ An important distinction to be drawn, in
particular, in the case of artifacts, is between functional and
Cerebral Cortex April 2008;18:740--751
doi:10.1093/cercor/bhm110
Advance Access publication July 9, 2007

The Author 2007. Published by Oxford University Press. All rights reserved.
For permissions, please e-mail: journals.permissions@oxfordjournals.org

action knowledge. Although these 2 types of knowledge have
been often treated as a unitary concept in the past, the
distinction between them has been recently supported by
neuropsychological evidence. In general, the use of an object
and the way it is manipulated do not bear any relationship. The
distinction between these two aspects is supported by the
classical neuropsychological distinction between semantic
deficits and apraxia. The latter has been associated with lesions
in those frontoparietal structures, which are also responsible for
visuomotor transformations from objects’ visual properties into
actions toward them, and in the retrieval of action knowledge
(Haaland et al. 2000). By definition, apraxic subjects should
show preserved object identification. On the other hand, there
are several reports on record of patients who show a profound
impairment of object knowledge but can nonetheless gesture
appropriately for objects they fail to identify, in association with
anterior, inferotemporal lesions (see, e.g., Sirigu et al. 1991;
Buxbaum et al. 1997; Lauro-Grotto et al. 1997; Hodges et al.
1999; Magnie et al. 1999).
The distinction between ‘‘what for’’ knowledge and ‘‘how’’
knowledge was further specified by Buxbaum and Saffran
(2002; see also Buxbaum et al. 2000). These authors studied 2
subpopulations of left hemisphere--lesioned patients and ob-
served that apraxic patients with frontoparietal lesions had
a specific loss of manipulation knowledge, which was associated
with impaired general knowledge about tools and body parts
but preserved functional knowledge. The opposite dissociation
was observed in a nonapraxics patients, with lesions confined in
the temporal lobe. Spatt et al. (2002) confirmed the presence of
a severe impairment in mechanical problem solving in cortico-
basal degeneration, which, however, in some cases may be
associated with defective conceptual knowledge, reflecting
the overlap of this condition with frontotemporal dementia
(Josephs et al. 2006). Moreover, Levy et al. (2004) reported
a significant correlation between the extent of anterolateral
temporal lesions and the severity of impaired semantic knowl-
edge (including functional features). The site of the lesion in
these patients is consistent with the results of a recent positron
emission tomography (PET) study, aiming to investigate the
cerebral regions recruited by the retrieval of knowledge re-
garding perceptual (structural and color) and functional aspects
of familiar objects (Kellenbach et al. 2005). In fact, the left
anterior middle/superior temporal regions and the temporal
pole were activated only by the retrieval of functional knowl-
edge, relative to both object structure and color.
The possible anatomical segregation ofmanipulative and func-
tional knowledge has been already investigated in 2 imaging
studies. The left posterior parietal cortex was reported to be
more strongly activated by the retrieval of manipulative, com-
pared with object’s function, knowledge both by Kellenbach
et al. (2003) and Boronat et al. (2005), using PET and epoch-
based functional magnetic resonance imaging (fMRI), respec-
tively. This region, together with its more occipital extension,
is widely considered to be where the interface between percep-
tually related and action-related processes occur (Rizzolatti
et al. 1997). The rostral portion of the inferior parietal lobule
and the ventral premotor cortex in the left hemisphere were
also activated in the same comparison in the former study.
These studies, however, failed to detect cerebral regions
specifically activated when retrieving functional knowledge.
In the present study, the cerebral organization of action and
functional semantic representations was investigated using
event-related modeling of single trials. Based on the above
reviewed neuropsychological evidence, we predicted stronger
anterior lateral temporal activations for the retrieval of func-
tional knowledge relative to action knowledge and stronger
frontoparietal activations in the opposite comparison.
Materials and Methods
Participants
Fifteen right-handed healthy monolingual native speakers of Italian (8
females and 7 males; mean age = 24.6 years; age range = 22--28 years)
with normal vision took part in the experiment. Handedness was
verified by means of the Edinburgh Inventory (Oldfield 1971). None of
them had a history of neurological or psychiatric disorders. Subjects
gave informed written consent to the experimental procedure, which
was approved by the local Ethics Committee.
Tasks and Experimental Procedure
A semantic-decision task was used, with 2 experimental conditions.
Subjects were presented with photographs showing pairs of manipu-
lable man-made objects and asked whether the 2 objects within each
pair were used with the same manipulation pattern (i.e., tongs and
potato squasher; vacuum cleaner and metal detector) or not (‘‘action
knowledge’’ condition, A), or in the same context, based on their
function (i.e., tongs and screwdriver; vacuum cleaner and carpet
beater) or not (‘‘functional knowledge’’ condition, F) (see Table 1 and
Fig. 1).
The study was composed by 8 scanning periods lasting 4 min 32 s
each. A blocked design for the presentation of the stimulus pairs was
used, with every period comprising one A and one F block condition,
each including 12 trials overall (see Fig. 1). In each trial, a pair of object
was presented. Every block started with a screen alerting the subject
(‘‘Ready!,’’ 1000 ms), followed by a screen displaying the instructions
(1500 ms), which were phrased as a question (‘‘Same manipulation?’’ or
‘‘Same context of use?’’ for A and F tasks, respectively). All verbal
instructions were presented in Italian. Instructions were followed by
a screen displaying the first pair of objects, which was visible for a 4-s
period during which subjects had to perform the task and prepare the
answer, based on the instructions and the presented objects. These
were followed by a white cross on black background, prompting
subjects to answer (Yes/No). Subjects were asked to give a vocal
response, which was recorded by means of a digital microphone lying
outside the scanner room and connected via a plastic tube in proximity
of the volunteers’ mouth. A vocal, rather than a manual, response was
used to reduce possible confoundings with the systems involved in the A
task. The appearance of the white cross and the onset of the next trial
were separated by a variable-length interval, allowing for implicit
modeling of the baseline. In order to optimize statistical efficiency,
interstimulus intervals between successive trials within a block were
presented in different (‘‘jittered’’) durations across trials (4.8, 7.2, and
10.1 s, in the proportion of 4:2:1) (Dale 1999). Any possible priming
effect within subjects was prevented for by requiring them to name all
the 48 stimuli prior to scanning. Stimulus pairs were viewed via a back-
projection screen located in front of the scanner and a mirror placed on
the head coil. Stimulus pairs were presented, and subjects’ answers and
experimental timing information were recorded, using the software
Presentation 9.13 (http://www.neurobs.com).
Table 1
An example of the object pairs in the 2 tasks
Task First object Second object Same
context?
Same
manipulation?
F Poultry shears Hand spiral beater Yes No
F Poultry shears Computer keyboard No No
A Poultry shears Tongs No Yes
A Poultry shears Stamp for postmark No No
Cerebral Cortex April 2008, V 18 N 4 741