Emerging Trends in The Social and Behavioral Sciences

Embodied Knowledge

Media

Part of Embodied Knowledge

Title

Embodied Knowledge

extracted text

Embodied Knowledge
DIANE PECHER and RENÉ ZEELENBERG

Abstract
In theories of grounded cognition, mental representations (concepts) share processing mechanisms with systems for perception and action. In this view, mental representations are simulations of embodied experiences. This view is supported by
empirical data showing that concepts, linguistic processing, and emotion processing
interact with perception and action. Key issues for further research are the question
how abstract concepts are grounded in sensory-motor processing, how language and
concepts are related, and the development of formal models.

INTRODUCTION
Experiences leave traces in people’s memory, forming mental concepts.
These mental concepts constitute the knowledge that people use to recognize objects, reason, make inferences, and thus shape human behavior. To
take a simple example, people have many experiences with objects such as
apples. These experiences leave traces in memory that allow the formation
of the concept of apples. This concept includes knowledge about perceptual
and motor features—what an apple looks like; how it feels, smells, and
tastes; and how one grasps an apple and takes a bite. When a person thinks
about the concept apple, for example, after reading the word apple, features
from the concept are activated and form the mental representation of apple.
In the past decade, more and more research has suggested that mental
concepts do not take the form of abstract symbols but are grounded in
perception and action.
FOUNDATIONAL RESEARCH
As Harnad (1990) noted, theories of mental representation need to solve the
grounding problem. If mental representations consist of arbitrary, abstract
symbols, a mechanism is needed that relates those symbols to real experiences. In most models, symbols get their meaning from their relations to
Emerging Trends in the Social and Behavioral Sciences. Edited by Robert Scott and Stephen Kosslyn.
© 2015 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.

1

2

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

other symbols. However, if symbols refer only to other symbols, they have
no intrinsic meaning but need an interpreter who knows the meaning of the
symbols. In this respect, understanding a set of symbols without grounding is like learning a language from a dictionary. One can look up a word,
but only finds other words. To understand the words, or symbols, one needs
to ground them in something familiar and meaningful. A good candidate
for such grounding is the sensory-motor experiences that people have in
the world. For example, the symbols for red, round, tart, juicy, can-be-bitten,
has-a-stem need to be linked to the perceptual and motoric experiences of
those properties in the real world. Once the concept apple is grounded in such
sensory-motor experiences it can be said to be meaningful.
The Perceptual Symbols Theory (Barsalou, 1999) was developed to solve
this grounding problem. Basically, Barsalou argues that the same processes
that are used for perception and action are also used for higher level cognition such as mental representations and language understanding. His theory
states that mental concepts are represented by perceptual symbols. Perceptual symbols are the neural states that underlie perception and action. People
learn these perceptual symbols through many perceptual and motor experiences with things in the world. Different sensory-motor experiences are connected via association areas similar to Damasio’s (1989) idea of convergence
zones. Association areas in modality-specific sensory-motor systems such as
the visual system and the motor system capture these experiences. Higher
order association areas integrate experiences from different sensory-motor
systems into multimodal experiences. Together, these association networks
at different levels activate the conceptual knowledge that is used during all
kinds of cognitive operations such as categorization and language comprehension. In order to create the mental representations needed for processing,
simulators reactivate partial experiences across different instances of a concept. Such simulations are activated top-down, from the higher level association areas all the way to the modality-specific sensory-motor systems. Importantly, in this view, the higher level association areas by themselves do not
actually represent any meaning. Rather, the sensory-motor systems provide
the content for mental representations. This is fundamentally different from
symbolic, amodal accounts of cognition, which assume that concepts can be
sufficiently represented by higher level abstract symbols.
EMPIRICAL SUPPORT FOR THE GROUNDING OF CONCEPTS
The Role of Perception. As briefly described above, grounded theories assume
that concepts share mechanisms with sensory-motor processing. In this view,
a concept, for example, apple, is represented by the simulation of potential
interactions with apples, such as seeing a round, red object, grasping it

Embodied Knowledge

3

with the hand and experiences of its firm feel, the sweet, tart, and juicy
experience of tasting it, and so on. This idea that concepts are supported
by sensory-motor systems makes the prediction that representational processes and perceptual processes should interact. This prediction has been
empirically tested in several ways.
In our laboratory, we have shown that modalities contribute to representation (Pecher, Zeelenberg, & Barsalou, 2003). During conceptual processing
switching between sensory modalities incurred a processing cost. Participants were slower and less accurate to verify that a concept has a particular property (e.g., apple-red) when the previous trial contained a property
from a different modality (e.g., airplane-noisy) than when it contained a property from the same modality (e.g., diamond-sparkling). The same effect was
obtained if, instead of verifying a concept-property pair, on the previous trial
participants indicated the location of a perceptual stimulus such as a burst
of noise or a light flash (Van Dantzig, Pecher, Zeelenberg, & Barsalou, 2008).
Such findings indicate that the mental simulation of a concept can be focused
on the relevant modality as if the observer is experiencing the concept in a
way that allows perception of the property. To verify that an airplane is noisy,
one must run a simulation of hearing an airplane, and to verify that a diamond sparkles, one must run a simulation of seeing a diamond. As is also
found with actual perception (Spence, Nicholls, & Driver, 2000), switching
between modalities incurs a cost because attention has to switch from one
modality to the other.
Another line of research has shown effects of mental representations on processing of visual stimuli (Stanfield & Zwaan, 2001; Zwaan, Madden, Yaxley,
& Aveyard, 2004; Zwaan, Stanfield, & Yaxley, 2002). The idea is that representation of an object and perception of that object might use (partly) overlapping perceptual features. Visual perception is facilitated by such overlapping
features. In several studies people read sentences in which an object’s orientation, shape, or motion was implied. For example, the sentence There was
an eagle in the sky implies an eagle with its wings out, whereas the sentence
There was an eagle in the nest implies an eagle with its wings folded in. Processing of a subsequent picture was facilitated when the relevant dimension
matched the one implied by the sentence (e.g., the shape of the wings) compared to when it mismatched. The effect of visual overlap does not only occur
between a currently activated concept and perception, but was also observed
when there was an hour-long interval between sentence reading and picture
processing (Pecher, Van Dantzig, Zwaan, & Zeelenberg, 2009). Thus, implied
visual features are not only represented during online language processing,
but also more available at longer delays. Both short and long-term effects
indicate that concepts not only contain perceptual features but also that the

4

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

particular features that are represented (e.g., orientation) are context dependent. Context-dependent activation of visual properties has been observed
even when no visual information is presented in the entire task. When participants are asked to list properties of concepts, they are more likely to name
properties that are visible from an implied perspective than properties that
are not visible (Wu & Barsalou, 2009). For example, seeds is named more frequently as a property of half a watermelon then as a property of a watermelon. In
our own laboratory, priming effects have been observed between words that
refer to objects with similar shapes (banjo-tennis racket), but only when the
experimental context has made shape relevant (Pecher, Zeelenberg, & Raaijmakers, 1998). Thus, perceptual features are relevant for concepts even in
linguistic contexts. This indicates that participants have visual representations of concepts.
The Role of Action. Equally important for concepts are actions. Glenberg
(1997) has proposed that the main function of concepts is to support interactions with the environment. Therefore, in his theory action is central to
mental representations. Two types of information are important for concepts.
First, objects have affordances. Affordances are actions that are possible
given the constraints of the environment and our bodies. For example, the
shape of an apple and the characteristics of our hands afford grasping the
apple with one hand. Research indicates that affordances are activated by
visual object information. Second, patterns of previous actions are stored
in memory. We have memories of bringing apples to our mouth and biting
them and in some cases these memories are activated when the object or a
reference to it is encountered. Concepts are formed by the combination of
current affordances and memories of previous actions. The affordances of
a particular apple are combined with memories of previous actions such as
biting the apple, giving the concept apple meaning. Support for this idea is
provided by interaction between concepts and action. For example, when
people judge sentences describing actions, their judgment is facilitated
when they have their hand shaped in a way that is appropriate for grasping
the object mentioned in the sentence (Klatzky, Pellegrino, McCloskey, &
Doherty, 1989) or when they move their hand in the direction that is implied
by the sentence (Glenberg & Kaschak, 2002). These and similar findings
have been obtained for different types of stimuli, such as object pictures,
words, and sentences. The effects occur in both directions, that is, actions
are affected by concepts and concepts are affected by actions. Moreover,
interactions between actions and concepts are even observed when the
concept is irrelevant for the action or vice versa (e.g., Bub & Masson, 2010).

Embodied Knowledge

5

Thus, these results strongly suggest that the motor system is involved in
representing concepts.
Flexibility. The view that concepts are grounded in sensory-motor systems
also implies that concepts are flexible. A major reason why concepts are flexible is that they are embedded in a context. People may encounter apples
in a variety of contexts: they may inspect apples in the store, looking for
exemplars without blemishes or they may bite an apple, feeling the firmness
of its skin and the juice running into their mouth, taste its sweetness and
tartness, and chew its flesh. There is evidence that linguistic context affects
which aspects of concepts are activated (Barsalou, 1993; Zeelenberg, Pecher,
Shiffrin, & Raaijmakers, 2003). Context can affect which sensory modalities
are relevant, and this in turn affects the strength of activation of features from
those modalities. Our studies on conceptual modality switching (Pecher et al.,
2003) suggest that attention can be focused on specific modalities if the concept is presented in the context of a feature from that modality. Properties
from a specific modality are more accessible if the prior context focuses on
that same modality than if it focuses on a different modality. These results
show that the context of the immediately preceding trial affects processing
on the current trial.
Related findings show that focusing on a specific modality can also have
long-term consequences. Activation of the visual features in a prior task
makes those features more available later on, possibly because of altered
representations. For example, we found that almost an hour after reading
the sentence chocolate is brown participant’s recognition memory was faster
for a black and white drawing of chocolate than after reading the sentence
chocolate is sweet (Pecher, Zanolie, & Zeelenberg, 2007). Because these
findings are obtained for representations that are based on linguistic input
they suggest that participants simulated a visual representation of the object
that has contextually relevant perceptual qualities. These properties are
then strengthened selectively, making them more available on subsequent
representations of the concept. Thus, concepts are flexible in the sense
that the availability of properties can vary. The representation at a specific
moment contains only a subset of all possible properties. Flexibility happens
at two time frames. The current context influences the current content of the
concept, thus having a short-term effect on representations. Information that
was activated is strengthened and as a result is more likely to be activated
again the next time the concept is represented. Thus, both the current context
and prior contexts affect the accessibility of concept features.

6

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

The Relation between Language and Concepts. Much of the evidence for mental
concepts as sensory-motor simulations comes from studies in which participants were processing language. Although language itself obviously has
visual, auditory, and motoric properties, the relation between sensory-motor
properties of an utterance and its meaning is quite arbitrary and abstract.
An important question is how language processing and conceptual processing interact. Rather than viewing these two types of processing as separate
modules or mechanisms, we can view them as aspects of the same experience. In this sense, linguistic experiences are sensory-motor experiences
that are connected via association areas at different levels, just like any other
sensory-motor experience. Such ideas have been worked out in formal models. For example, Plaut (2002) developed a connectionist model that consists
of modality-specific and language input and output subsystems and a central layer connecting all input and output systems. His simulations suggest
that both within-modality and cross-modality connections are important to
represent conceptual structure. A related idea is that of the semantic hub (Patterson, Nestor, & Rogers, 2007). The semantic hub, supposedly located in the
anterior temporal pole, is a single area that connects different sensory-motor
and language processing areas and has some representational power. Thus,
in all these proposals language and sensory-motor activations are part of the
same network that supports both linguistic processing and conceptual processing.
Such a network explains how concepts are formed and how concepts and
language are related. Experiences with concepts are often accompanied by
language. For example, one may see the word apple together with a picture of
an apple, or hear the word apple followed by the taste of an apple. Experiences
in different modalities, both directly with the object and with the words that
are related to the concept become interconnected. This way, coherent concepts such as apple are formed. Because words and concepts are related, language activates the concept and sensory-motor activations activate language.
An important role for language and the association areas or semantic hub is
the formation of categories. Superordinate categories such as furniture often
consist of exemplars that have little perceptual similarity with each other and
might even share perceptual similarity with exemplars from completely different categories. Language helps to bind experiences with exemplars of the
same category even if they do not share perceptual experience and helps to
distinguish exemplars that share perceptual features but are from different
categories.
Language might also play a role as a shortcut in tasks that do not require
deep conceptual processing (Simmons, Hamann, Harenski, Hu, & Barsalou,
2008). Word associations are produced faster than responses that require perceptual simulations. Studies show that when a task can be performed on the

Embodied Knowledge

7

basis of simple word associations, the role of sensory-motor processing is
smaller than when word associations are not enough to perform a task. This
suggests that when people are giving fast responses they rely on word associations and do not fully activate concepts. Word co-occurrence norms also
show, however, that sensory-motor variables are captured in linguistic utterances. Although some researchers have interpreted this finding as evidence
for symbolic representations of concepts (Louwerse & Jeuniaux, 2008), we
think it is more likely that concepts are represented by sensory-motor activations and that these activations are reflected in their utterances.
Emotion. In addition to language processing, the role of sensory-motor activations for conceptual processing can also be clearly seen in how people
process emotion. The valence of a concept is the degree to which people have
positive or negative feelings about it. For example, people may have positive
feelings about kittens and negative feelings about snakes. Such feelings are
associated with approach and avoidance actions. If you like kittens, you have
a tendency to bring them closer to yourself, for example, by taking a step
in their direction or by picking one up and bringing it closer to your body.
Conversely, if you do not like snakes you have a tendency to increase the
distance between your body and the snake. In the laboratory, this is investigated by showing pictures or words with emotional valence and asking
participants to respond to the stimuli by moving a lever toward or away from
themselves, mimicking approach and avoidance actions respectively. People
find it easier to make approach responses when the stimulus has positive
valence than when it has negative valence and to make avoidance responses
when it has negative valence than when it has positive valence (Chen &
Bargh, 1999). The relation between valence and approach/avoidance actions
also works in the opposite direction. People tend to like meaningless stimuli
more after performing an approach action than after performing an avoidance reaction. Another important and related finding is mimicry. People have
a tendency to mimic other people’s facial expressions or bodily postures.
Researchers have suggested that mimicry underlies understanding of and
empathy with other people’s emotions (Niedenthal, Barsalou, Winkielman,
Krauth-Gruber, & Ric, 2005). People may use feedback from their own expression to recognize other people’s feelings. Mimicry is stronger between people who like each other than between people who do not like each other,
and when people are prevented from using mimicry their performance in
emotion recognition tasks drops. Thus, the close relation between emotion
understanding and emotional actions suggest that action is fundamental to
emotional concepts.

8

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

RECENT DEVELOPMENTS: THE GROUNDING OF ABSTRACT CONCEPTS
The grounded cognition approach has been criticized for dealing mainly
with the mental representations of concrete objects and actions. For concrete
objects and actions it is relatively easy to see how sensory-motor systems
might be involved in their representations. As we discussed above, there is
much evidence that conceptual processing and sensory-motor processing
interact, providing support for the idea that perception, action, and mental
representations share processing resources. In order to claim that cognition
is grounded in perception and action, however, we need to show that such
interactions exist not only for concrete objects such as apples but also for
more abstract concepts such as truth and problem solving.
Several ideas have been put forward to solve the grounding problem for
abstract concepts (Pecher, Boot, & Van Dantzig, 2011). The most thoroughly
investigated idea has its origin in cognitive linguistics, the Conceptual
Metaphor Theory (Lakoff & Johnson, 1980). On this account, people use
metaphors and image schemas to ground abstract concepts in sensory-motor
experiences. In language, people often use concrete domains to talk about
abstract concepts. For example, people talk about solving a problem in terms
of a journey that involves traveling from a starting point (the problem
situation) to a destination (the solution) along a path (the method that is
used to solve the problem). This can be seen in expressions such as to have
something get in one’s way or to reach the solution. People have very concrete
experiences with journeys, full of rich sensory-motor details. These concrete
experiences are mapped on the abstract situation of solving the problem,
giving it concrete structure. This account explains how abstract concepts
are grounded in sensory-motor systems because the mapping is based on
concrete physical experiences.
Starting with this linguistic evidence, cognitive scientists have further
investigated the psychological basis of Conceptual Metaphor Theory.
Although metaphors are observed in language, the theory claims that
metaphorical language is the expression of the underlying metaphorical
concepts. Thus, metaphors are not merely a linguistic phenomenon, but are
fundamental to conceptual processing. The theory distinguishes between
primary metaphors, or image schemas, and secondary metaphors. Especially
image schemas are relevant for grounded cognition. Image schemas are
basic patterns of embodied experience, such as up-down, inside-outside, and
balance. They are learned through multimodal experiences. The resulting
image schemas are analogue representations of embodied experiences in
space but they are not modality specific. Rather, they are more abstract
than direct sensory-motor experiences, and in this way can form a bridge
between direct physical interactions and abstract concepts.

Embodied Knowledge

9

Evidence for Conceptual Metaphor Theory is most convincing when image
schemas are activated in the absence of any linguistic reference to the image
schema. Studies have shown that mental representations of abstract concepts
result in activation of concrete image schemas. For example, researchers
have investigated whether the abstract concept power activates an up-down
image schema (Schubert, 2005). When people talk about power, they often
refer to powerful people as being higher than powerless people. Consistent
with this mapping, participants shifted their visual-spatial attention to
higher locations after reading a word referring to a person with much power
(e.g., president) than after reading a word referring to a person with little
power (e.g., servant). As a result, participants were better at identifying
an unrelated visual stimulus in congruent locations than in incongruent
locations (Zanolie et al., 2012). Similar findings have been reported on the
activation of spatial proximity for the concept similarity, up-down for quantity,
containment for category membership, spatial movement for time, and left-right
for valence (see Pecher et al., 2011, for an overview). All these studies provide
evidence for a role of image schemas for mental representations of abstract
concepts in contexts that contain no direct reference to the metaphorical
mapping.
Image schemas thus seem to play a role in de representation of abstract concepts. The question is to what extent Conceptual Metaphor Theory can fully
explain abstract concepts. Image schemas are very basic patterns, representing a single dimension with limited values. For example, the up-down image
schema represents only a concept’s vertical relative position. This alone is
not sufficient to represent the concept power. Rather, vertical position is an
embodied representation of relative differences in power between people. It
does not represent what it means to have power over someone, such as having the ability to make someone do something. A related issue is that there is
no one-to-one mapping between concepts and image schemas. For example,
many metaphors can be used for the concept love, such as journey, proximity,
and containment. Similarly, a single metaphor can be used for many concepts.
For example, the up-down metaphor can be used for power, quantity, valence,
and so on. This also illustrates that image schemas have only limited representational power. In other words, an image schema represents one property
of an abstract concept, and additional information is needed to represent the
full meaning of a concept.
Such additional details might be provided by situated concepts. In this
view, concepts develop from entire situations. A concept is a collection
of experiences that are combined into a representation. For example, the
concept power is based on many situated experiences that a person has with
power, and the representation of power likely occurs in a situated context.

10

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Abstract concepts often derive much of their meaning from an entire situation, more so than concrete concepts. Because situations have sensory-motor
properties, this proposal can also explain how abstract concepts might be
grounded. The representational power of such concept can be based entirely
on collections of experiences. This idea was recently further developed for
the emotion concepts fear and anger (Wilson-Mendenhall, Barrett, Simmons,
& Barsalou, 2011). Exemplar models of categorization (Hintzman, 1986;
Nosofsky, 1986) explain how individual episodes of experience are stored.
Abstraction occurs when a cue activates several experiences. These different
experiences are then combined into some sort of summary representation
for further processing. Thus, there is no need for the storage of abstract
representations, instead, abstractions are created as a response to a cue.
Another approach to grounding abstract concepts is to combine embodied
experiences with linguistic relations (Andrews, Vigliocco, & Vinson, 2009).
On this account, concepts are acquired through these two sources. Embodied
experiences provide sensory-motor grounding, but this may not be available
for abstract concepts. Knowledge is also acquired through language, however, and the availability of linguistic information is independent of how
concrete or abstract the concept is. They developed a model in which the
two sources are combined. In this model, abstract concepts are represented
as linguistic items that are related to other linguistic items and that, by inferences, are also related to sensory-motor information. Thus, abstract concepts
are grounded, albeit indirectly, in sensory-motor activations.
KEY ISSUES FOR FURTHER RESEARCH
Clearly, the question of how abstract concepts are grounded needs further
research. Although there is quite some evidence that image schemas are
important for abstract concepts, image schemas may be of limited value
for representations because they only represent certain properties. Richer
representations might consist of situational properties. Situations play a
role for all concepts, and perhaps even more so for abstract concepts than
for concrete concepts. In fact, it seems impossible to represent an abstract
concept such as power without the context. Thus, a theory of abstract
concepts should take situations into account. Situations should likely be
considered part of the concept rather than just background. For example,
experiments with problem solving tasks have shown that concrete details of
the situation are very important, even if they are irrelevant for the proper
solution of the problem. Research on transfer in problem solving tasks
such as in physics has shown that participants are more likely to use a
previously learned solution if problems share superficial properties, even
if the underlying structure is very different (Goldstone & Sakamoto, 2003).

Embodied Knowledge

11

One might argue that solving physics problems is a very difficult task for
novices. A question for future research is whether the same is true for more
automatic processes such as mental representation.
A second important issue is the role of language. First, language might
be important for the acquisition and perhaps also the representation of
abstract concepts. In Andrews et al.’s (2009) model concrete concepts are
grounded directly in sensory-motor experiences but are also represented by
the pattern of linguistic contexts in which they appear. Abstract concepts are
mostly represented by the linguistic contexts. One might wonder to what
extent representations can still be considered grounded in sensory-motor
processing if the grounding is indirect through other linguistic symbols.
Second, other researchers have even argued that linguistic data such as
word co-occurrences show effects of sensory-motor aspects of concepts. For
example, concepts that are perceptually similar occur in linguistically similar
contexts. Thus, linguistic relations might be sufficient for the representation
of meaning. One should take into account, however, that linguistic data
are produced by humans who already have rich concepts. Models of word
co-occurrences have as input texts such as internet news groups or textbooks
that are used in schools. All these texts are produced by people, and thus can
be expected to reflect the conceptual structure of the person producing the
test. Thus, rather than concluding that conceptual representation is based
on linguistic symbols, one could say they are expressions of underlying
concepts and have no bearing on the nature of those underlying concepts.
Attempts to use linguistic output to investigate the nature of the underlying
representations should therefore be regarded with some care.
A third issue is the need for formal models of grounded cognition. The
literature seems to be replete with claims that the grounded cognition
framework makes certain predictions but these claims are based on verbal
theories rather than detailed formal models. Formal models can give insight
into the feasibility of the approach. So far, few attempts have been made at
formal modeling. One important obstacle might be the operationalization of
sensory-motor input. If the input to a model is derived from verbal materials, such as feature production norms, one has to be aware that this is only
an approximation to actual sensory-motor representation. Computational
models of human cognition have played an important role in understanding
human behavior and may also help in better understanding how meaning is
represented.
CONCLUSION
To conclude, as many researchers from fields such as philosophy, linguistics,
and psychology have remarked, concepts need to be grounded. In the past

12

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

15 years, several ideas have been put forward on how concepts might be
grounded in sensory-motor processing. Inspired by these ideas, researchers
have obtained support for such ideas. Now, however, it seems we are reaching the boundaries. Some of these boundaries are mentioned above. If we
can jump these boundaries, the field will make further progress toward a
grounded view of cognition.
REFERENCES
Andrews, M., Vigliocco, G., & Vinson, D. (2009). Integrating experiential and distributional data to learn semantic representations. Psychological Review, 116, 463–498.
doi:10.1037/a0016261
Barsalou, L. W. (1993). Flexibility, structure, and linguistic vagary in concepts: Manifestations of a compositional system of perceptual symbols. In A. F. Collins, S. E.
Gathercole, M. A. Conway & P. E. Morris (Eds.), Theories of memory (pp. 29–101).
Hove, England: Erlbaum.
Barsalou, L. W. (1999). Perceptual symbol systems. Behavioral and Brain Sciences, 22,
577–660. doi:10.1017/S0140525X99002149
Bub, D. N., & Masson, M. E. J. (2010). Grasping beer mugs: On the dynamics of alignment effects induced by handled objects. Journal of Experimental Psychology: Human
Perception and Performance, 36, 341–358. doi:10.1037/a0017606
Chen, M., & Bargh, J. A. (1999). Consequences of automatic evaluation: Immediate
behavioral predispositions to approach or avoid the stimulus. Personality and Social
Psychology Bulletin, 25, 215–224.
Damasio, A. R. (1989). Time-locked multiregional retroactivation: A systems-level
proposal for the neural substrates of recall and recognition. Cognition, 33, 25–62.
doi:10.1016/0010-0277(89)90005-X
Glenberg, A. M. (1997). What memory is for. Behavioral and Brain Sciences, 20, 1–55.
doi:10.1017/S0140525X97000010
Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in action. Psychonomic
Bulletin & Review, 9, 558–565. doi:10.3758/BF03196313
Goldstone, R. L., & Sakamoto, Y. (2003). The transfer of abstract principles governing complex adaptive systems. Cognitive Psychology, 46, 414–466. doi:10.1016/
S0010-0285(02)00519-4
Harnad, S. (1990). The symbol grounding problem. Physica D, 42, 335–346.
doi:10.1016/0167-2789(90)90087-6
Hintzman, D. L. (1986). Schema abstraction in a multiple-trace memory model. Psychological Review, 93, 411–428. doi:10.1037//0033-295X.93.4.411
Klatzky, R. L., Pellegrino, J. W., McCloskey, B. P., & Doherty, S. (1989). Can you
squeeze a tomato? The role of motor representations in semantic sensibility
judgments. Journal of Memory and Language, 28, 56–77. doi:10.1016/0749-596X(89)
90028-4
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago, IL: Chicago University
Press.

Embodied Knowledge

13

Louwerse, M. M., & Jeuniaux, P. (2008). Language comprehension is both embodied
and symbolic. In M. De Vega, A. M. Glenberg & A. C. Graesser (Eds.), Embodiment and meaning: A debate (pp. 309–326). Oxford, England: Oxford University
Press.
Niedenthal, P. M., Barsalou, L. W., Winkielman, P., Krauth-Gruber, S., & Ric, F. (2005).
Embodiment in attitudes, social perception, and emotion. Personality and Social
Psychology Review, 9, 184–211. doi:10.1207/s15327957pspr0903_1
Nosofsky, R. M. (1986). Attention, similarity, and the identification-categorization
relationship. Journal of Experimental Psychology: General, 115, 39–57. doi:10.1037/
0096-3445.115.1.39
Patterson, K., Nestor, P. J., & Rogers, T. T. (2007). Where do you know what you
know? The representation of semantic knowledge in the human brain. Nature
Reviews: Neuroscience, 8, 976–987. doi:10.1038/nrn2277
Pecher, D., Boot, I., & Van Dantzig, S. (2011). Abstract concepts: Sensory-motor
grounding, metaphors, and beyond. In B. H. Ross (Ed.), The psychology of learning and motivation (pp. 217–248). Burlington, MA: Academic Press 10.1016/B9780-12-385527-5.00007-3
Pecher, D., Van Dantzig, S., Zwaan, R. A., & Zeelenberg, R. (2009). Language comprehenders retain implied shape and orientation of objects. Quarterly Journal of
Experimental Psychology, 62, 1108–1114. doi:10.1080/17470210802633255
Pecher, D., Zanolie, K., & Zeelenberg, R. (2007). Verifying visual properties in sentence verification facilitates picture recognition memory. Experimental Psychology,
54, 173–179. doi:10.1027/1618-3169.54.3.173
Pecher, D., Zeelenberg, R., & Barsalou, L. W. (2003). Verifying different-modality
properties for concepts produces switching costs. Psychological Science, 14, 119–124.
doi:10.1111/1467-9280.t01-1-01429
Pecher, D., Zeelenberg, R., & Raaijmakers, J. G. W. (1998). Does pizza prime coin?
Perceptual priming in lexical decision and pronunciation. Journal of Memory and
Language, 38, 401–418. doi:10.1006/jmla.1997.2557
Plaut, D. C. (2002). Graded modality specific specialisation in semantics: A computational account of optic aphasia. Cognitive Neuropsychology, 19, 603–639.
doi:10.1080/02643290244000112
Schubert, T. W. (2005). Your highness: Vertical positions as perceptual symbols of
power. Journal of Personality and Social Psychology, 89, 1–21. doi:10.1037/00223514.89.1.1
Simmons, W. K., Hamann, S. B., Harenski, C. L., Hu, X. P., & Barsalou, L. W. (2008).
fMRI evidence for word association and situated simulation in conceptual processing. Journal of Physiology, Paris, 102, 106–119. doi:10.1016/j.jphysparis.2008.03.014
Spence, C., Nicholls, M. E. R., & Driver, J. (2000). The cost of expecting events in the
wrong sensory modality. Perception & Psychophysics, 63, 330–336.
Stanfield, R. A., & Zwaan, R. A. (2001). The effect of implied orientation derived
from verbal context on picture recognition. Psychological Science, 12, 153–156.
doi:10.1111/1467-9280.00326
Van Dantzig, S., Pecher, D., Zeelenberg, R., & Barsalou, L. W. (2008). Perceptual processing affects conceptual processing. Cognitive Science, 32, 579–590. doi:10.1080/
03640210802035365

14

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Wilson-Mendenhall, C. D., Barrett, L. F., Simmons, W. K., & Barsalou, L. W. (2011).
Grounding emotion in situated conceptualization. Neuropsychologia, 49, 1105–1127.
doi:10.1016/j.neuropsychologia.2010.12.032
Wu, L., & Barsalou, L. W. (2009). Perceptual simulation in conceptual combination:
Evidence from property generation. Acta Psychologica, 132, 173–189. doi:10.1016/
j.actpsy.2009.02.002
Zanolie, K., Van Dantzig, S., Boot, I., Wijnen, J., Schubert, T. W., Giessner, S.
R., & Pecher, D. (2012). Mighty metaphors: Behavioral and ERP evidence that
power shifts attention on a vertical dimension. Brain and Cognition, 78, 50–58.
doi:10.1016/j.bandc.2011.10.006
Zeelenberg, R., Pecher, D., Shiffrin, R. M., & Raaijmakers, J. G. W. (2003). Semantic
context effects and priming in word association. Psychonomic Bulletin & Review, 10,
653–660.
Zwaan, R. A., Madden, C. J., Yaxley, R. H., & Aveyard, M. E. (2004). Moving
words: Dynamic representations in language comprehension. Cognitive Science, 28,
611–619. doi:10.1207/s15516709cog2804_5
Zwaan, R. A., Stanfield, R. A., & Yaxley, R. H. (2002). Language comprehenders mentally represent the shape of objects. Psychological Science, 13, 168–171. doi:10.1111/
1467-9280.00430

DIANE PECHER SHORT BIOGRAPHY
Diane Pecher is an Associate Professor at the Psychology Department of the
Erasmus University Rotterdam, the Netherlands. She holds a PhD from the
University of Amsterdam. Her research concerns conceptual memory and its
relation to language, perception, motor actions, and other types of memory.
RENÉ ZEELENBERG SHORT BIOGRAPHY
René Zeelenberg is an Associate Professor at the Psychology Department of
the Erasmus University Rotterdam, the Netherlands. He holds a PhD from
the University of Amsterdam. Rene spent several years at various US Universities as a Postdoc (Indiana University) and Visiting Scholar (Emory University, UCSD). His research concerns long-term memory, conceptual knowledge, and the influences of emotion on cognition.
Web sites:
www.memorylab.eu
eur.academia.edu/DianePecher
eur.academia.edu/ReneZeelenberg
www.brain-cognition.eu

Embodied Knowledge

15

RELATED ESSAYS
Theory of Mind and Behavior (Psychology), Amanda C. Brandone
Neighborhoods and Cognitive Development (Psychology), Jondou Chen and
Jeanne Brooks-Gunn
Language, Perspective, and Memory (Psychology), Rachel A. Ryskin et al.
Stereotype Content (Sociology), Beatrice H. Capestany and Lasana T. Harris
Culture and Cognition (Sociology), Karen A. Cerulo
The Inherence Heuristic: Generating Everyday Explanations (Psychology),
Andrei Cimpian
Cultural Differences in Emotions (Psychology), Jozefien De Leersnyder et al.
Micro-Cultures (Sociology), Gary Alan Fine
Economics and Culture (Economics), Gérard Roland
Language and Thought (Psychology), Susan Goldin-Meadow
Cultural Neuroscience: Connecting Culture, Brain, and Genes (Psychology),
Shinobu Kitayama and Sarah Huff
Resource Limitations in Visual Cognition (Psychology), Brandon M. Liverence and Steven L. Franconeri
Immigrant Sociocultural Adaptation, Identification, and Belonging (Sociology), Sarah J. Mahler
Concepts and Semantic Memory (Psychology), Barbara C. Malt
Emerging Trends in Culture and Concepts (Psychology), Bethany Ojalehto
and Douglas Medin
Neural and Cognitive Plasticity (Psychology), Eduardo Mercado III
A Bio-Social-Cultural Approach to Early Cognitive Development: Entering
the Community of Minds (Psychology), Katherine Nelson
Social Classification (Sociology), Elizabeth G. Pontikes
Attention and Perception (Psychology), Ronald A. Rensink
Class, Cognition, and Face-to-Face Interaction (Sociology), Lauren A. Rivera
Speech Perception (Psychology), Athena Vouloumanos
Theory of Mind (Psychology), Henry Wellman



Embodied Knowledge
DIANE PECHER and RENÉ ZEELENBERG

Abstract
In theories of grounded cognition, mental representations (concepts) share processing mechanisms with systems for perception and action. In this view, mental representations are simulations of embodied experiences. This view is supported by
empirical data showing that concepts, linguistic processing, and emotion processing
interact with perception and action. Key issues for further research are the question
how abstract concepts are grounded in sensory-motor processing, how language and
concepts are related, and the development of formal models.

INTRODUCTION
Experiences leave traces in people’s memory, forming mental concepts.
These mental concepts constitute the knowledge that people use to recognize objects, reason, make inferences, and thus shape human behavior. To
take a simple example, people have many experiences with objects such as
apples. These experiences leave traces in memory that allow the formation
of the concept of apples. This concept includes knowledge about perceptual
and motor features—what an apple looks like; how it feels, smells, and
tastes; and how one grasps an apple and takes a bite. When a person thinks
about the concept apple, for example, after reading the word apple, features
from the concept are activated and form the mental representation of apple.
In the past decade, more and more research has suggested that mental
concepts do not take the form of abstract symbols but are grounded in
perception and action.
FOUNDATIONAL RESEARCH
As Harnad (1990) noted, theories of mental representation need to solve the
grounding problem. If mental representations consist of arbitrary, abstract
symbols, a mechanism is needed that relates those symbols to real experiences. In most models, symbols get their meaning from their relations to
Emerging Trends in the Social and Behavioral Sciences. Edited by Robert Scott and Stephen Kosslyn.
© 2015 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.

1

2

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

other symbols. However, if symbols refer only to other symbols, they have
no intrinsic meaning but need an interpreter who knows the meaning of the
symbols. In this respect, understanding a set of symbols without grounding is like learning a language from a dictionary. One can look up a word,
but only finds other words. To understand the words, or symbols, one needs
to ground them in something familiar and meaningful. A good candidate
for such grounding is the sensory-motor experiences that people have in
the world. For example, the symbols for red, round, tart, juicy, can-be-bitten,
has-a-stem need to be linked to the perceptual and motoric experiences of
those properties in the real world. Once the concept apple is grounded in such
sensory-motor experiences it can be said to be meaningful.
The Perceptual Symbols Theory (Barsalou, 1999) was developed to solve
this grounding problem. Basically, Barsalou argues that the same processes
that are used for perception and action are also used for higher level cognition such as mental representations and language understanding. His theory
states that mental concepts are represented by perceptual symbols. Perceptual symbols are the neural states that underlie perception and action. People
learn these perceptual symbols through many perceptual and motor experiences with things in the world. Different sensory-motor experiences are connected via association areas similar to Damasio’s (1989) idea of convergence
zones. Association areas in modality-specific sensory-motor systems such as
the visual system and the motor system capture these experiences. Higher
order association areas integrate experiences from different sensory-motor
systems into multimodal experiences. Together, these association networks
at different levels activate the conceptual knowledge that is used during all
kinds of cognitive operations such as categorization and language comprehension. In order to create the mental representations needed for processing,
simulators reactivate partial experiences across different instances of a concept. Such simulations are activated top-down, from the higher level association areas all the way to the modality-specific sensory-motor systems. Importantly, in this view, the higher level association areas by themselves do not
actually represent any meaning. Rather, the sensory-motor systems provide
the content for mental representations. This is fundamentally different from
symbolic, amodal accounts of cognition, which assume that concepts can be
sufficiently represented by higher level abstract symbols.
EMPIRICAL SUPPORT FOR THE GROUNDING OF CONCEPTS
The Role of Perception. As briefly described above, grounded theories assume
that concepts share mechanisms with sensory-motor processing. In this view,
a concept, for example, apple, is represented by the simulation of potential
interactions with apples, such as seeing a round, red object, grasping it

Embodied Knowledge

3

with the hand and experiences of its firm feel, the sweet, tart, and juicy
experience of tasting it, and so on. This idea that concepts are supported
by sensory-motor systems makes the prediction that representational processes and perceptual processes should interact. This prediction has been
empirically tested in several ways.
In our laboratory, we have shown that modalities contribute to representation (Pecher, Zeelenberg, & Barsalou, 2003). During conceptual processing
switching between sensory modalities incurred a processing cost. Participants were slower and less accurate to verify that a concept has a particular property (e.g., apple-red) when the previous trial contained a property
from a different modality (e.g., airplane-noisy) than when it contained a property from the same modality (e.g., diamond-sparkling). The same effect was
obtained if, instead of verifying a concept-property pair, on the previous trial
participants indicated the location of a perceptual stimulus such as a burst
of noise or a light flash (Van Dantzig, Pecher, Zeelenberg, & Barsalou, 2008).
Such findings indicate that the mental simulation of a concept can be focused
on the relevant modality as if the observer is experiencing the concept in a
way that allows perception of the property. To verify that an airplane is noisy,
one must run a simulation of hearing an airplane, and to verify that a diamond sparkles, one must run a simulation of seeing a diamond. As is also
found with actual perception (Spence, Nicholls, & Driver, 2000), switching
between modalities incurs a cost because attention has to switch from one
modality to the other.
Another line of research has shown effects of mental representations on processing of visual stimuli (Stanfield & Zwaan, 2001; Zwaan, Madden, Yaxley,
& Aveyard, 2004; Zwaan, Stanfield, & Yaxley, 2002). The idea is that representation of an object and perception of that object might use (partly) overlapping perceptual features. Visual perception is facilitated by such overlapping
features. In several studies people read sentences in which an object’s orientation, shape, or motion was implied. For example, the sentence There was
an eagle in the sky implies an eagle with its wings out, whereas the sentence
There was an eagle in the nest implies an eagle with its wings folded in. Processing of a subsequent picture was facilitated when the relevant dimension
matched the one implied by the sentence (e.g., the shape of the wings) compared to when it mismatched. The effect of visual overlap does not only occur
between a currently activated concept and perception, but was also observed
when there was an hour-long interval between sentence reading and picture
processing (Pecher, Van Dantzig, Zwaan, & Zeelenberg, 2009). Thus, implied
visual features are not only represented during online language processing,
but also more available at longer delays. Both short and long-term effects
indicate that concepts not only contain perceptual features but also that the

4

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

particular features that are represented (e.g., orientation) are context dependent. Context-dependent activation of visual properties has been observed
even when no visual information is presented in the entire task. When participants are asked to list properties of concepts, they are more likely to name
properties that are visible from an implied perspective than properties that
are not visible (Wu & Barsalou, 2009). For example, seeds is named more frequently as a property of half a watermelon then as a property of a watermelon. In
our own laboratory, priming effects have been observed between words that
refer to objects with similar shapes (banjo-tennis racket), but only when the
experimental context has made shape relevant (Pecher, Zeelenberg, & Raaijmakers, 1998). Thus, perceptual features are relevant for concepts even in
linguistic contexts. This indicates that participants have visual representations of concepts.
The Role of Action. Equally important for concepts are actions. Glenberg
(1997) has proposed that the main function of concepts is to support interactions with the environment. Therefore, in his theory action is central to
mental representations. Two types of information are important for concepts.
First, objects have affordances. Affordances are actions that are possible
given the constraints of the environment and our bodies. For example, the
shape of an apple and the characteristics of our hands afford grasping the
apple with one hand. Research indicates that affordances are activated by
visual object information. Second, patterns of previous actions are stored
in memory. We have memories of bringing apples to our mouth and biting
them and in some cases these memories are activated when the object or a
reference to it is encountered. Concepts are formed by the combination of
current affordances and memories of previous actions. The affordances of
a particular apple are combined with memories of previous actions such as
biting the apple, giving the concept apple meaning. Support for this idea is
provided by interaction between concepts and action. For example, when
people judge sentences describing actions, their judgment is facilitated
when they have their hand shaped in a way that is appropriate for grasping
the object mentioned in the sentence (Klatzky, Pellegrino, McCloskey, &
Doherty, 1989) or when they move their hand in the direction that is implied
by the sentence (Glenberg & Kaschak, 2002). These and similar findings
have been obtained for different types of stimuli, such as object pictures,
words, and sentences. The effects occur in both directions, that is, actions
are affected by concepts and concepts are affected by actions. Moreover,
interactions between actions and concepts are even observed when the
concept is irrelevant for the action or vice versa (e.g., Bub & Masson, 2010).

Embodied Knowledge

5

Thus, these results strongly suggest that the motor system is involved in
representing concepts.
Flexibility. The view that concepts are grounded in sensory-motor systems
also implies that concepts are flexible. A major reason why concepts are flexible is that they are embedded in a context. People may encounter apples
in a variety of contexts: they may inspect apples in the store, looking for
exemplars without blemishes or they may bite an apple, feeling the firmness
of its skin and the juice running into their mouth, taste its sweetness and
tartness, and chew its flesh. There is evidence that linguistic context affects
which aspects of concepts are activated (Barsalou, 1993; Zeelenberg, Pecher,
Shiffrin, & Raaijmakers, 2003). Context can affect which sensory modalities
are relevant, and this in turn affects the strength of activation of features from
those modalities. Our studies on conceptual modality switching (Pecher et al.,
2003) suggest that attention can be focused on specific modalities if the concept is presented in the context of a feature from that modality. Properties
from a specific modality are more accessible if the prior context focuses on
that same modality than if it focuses on a different modality. These results
show that the context of the immediately preceding trial affects processing
on the current trial.
Related findings show that focusing on a specific modality can also have
long-term consequences. Activation of the visual features in a prior task
makes those features more available later on, possibly because of altered
representations. For example, we found that almost an hour after reading
the sentence chocolate is brown participant’s recognition memory was faster
for a black and white drawing of chocolate than after reading the sentence
chocolate is sweet (Pecher, Zanolie, & Zeelenberg, 2007). Because these
findings are obtained for representations that are based on linguistic input
they suggest that participants simulated a visual representation of the object
that has contextually relevant perceptual qualities. These properties are
then strengthened selectively, making them more available on subsequent
representations of the concept. Thus, concepts are flexible in the sense
that the availability of properties can vary. The representation at a specific
moment contains only a subset of all possible properties. Flexibility happens
at two time frames. The current context influences the current content of the
concept, thus having a short-term effect on representations. Information that
was activated is strengthened and as a result is more likely to be activated
again the next time the concept is represented. Thus, both the current context
and prior contexts affect the accessibility of concept features.

6

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

The Relation between Language and Concepts. Much of the evidence for mental
concepts as sensory-motor simulations comes from studies in which participants were processing language. Although language itself obviously has
visual, auditory, and motoric properties, the relation between sensory-motor
properties of an utterance and its meaning is quite arbitrary and abstract.
An important question is how language processing and conceptual processing interact. Rather than viewing these two types of processing as separate
modules or mechanisms, we can view them as aspects of the same experience. In this sense, linguistic experiences are sensory-motor experiences
that are connected via association areas at different levels, just like any other
sensory-motor experience. Such ideas have been worked out in formal models. For example, Plaut (2002) developed a connectionist model that consists
of modality-specific and language input and output subsystems and a central layer connecting all input and output systems. His simulations suggest
that both within-modality and cross-modality connections are important to
represent conceptual structure. A related idea is that of the semantic hub (Patterson, Nestor, & Rogers, 2007). The semantic hub, supposedly located in the
anterior temporal pole, is a single area that connects different sensory-motor
and language processing areas and has some representational power. Thus,
in all these proposals language and sensory-motor activations are part of the
same network that supports both linguistic processing and conceptual processing.
Such a network explains how concepts are formed and how concepts and
language are related. Experiences with concepts are often accompanied by
language. For example, one may see the word apple together with a picture of
an apple, or hear the word apple followed by the taste of an apple. Experiences
in different modalities, both directly with the object and with the words that
are related to the concept become interconnected. This way, coherent concepts such as apple are formed. Because words and concepts are related, language activates the concept and sensory-motor activations activate language.
An important role for language and the association areas or semantic hub is
the formation of categories. Superordinate categories such as furniture often
consist of exemplars that have little perceptual similarity with each other and
might even share perceptual similarity with exemplars from completely different categories. Language helps to bind experiences with exemplars of the
same category even if they do not share perceptual experience and helps to
distinguish exemplars that share perceptual features but are from different
categories.
Language might also play a role as a shortcut in tasks that do not require
deep conceptual processing (Simmons, Hamann, Harenski, Hu, & Barsalou,
2008). Word associations are produced faster than responses that require perceptual simulations. Studies show that when a task can be performed on the

Embodied Knowledge

7

basis of simple word associations, the role of sensory-motor processing is
smaller than when word associations are not enough to perform a task. This
suggests that when people are giving fast responses they rely on word associations and do not fully activate concepts. Word co-occurrence norms also
show, however, that sensory-motor variables are captured in linguistic utterances. Although some researchers have interpreted this finding as evidence
for symbolic representations of concepts (Louwerse & Jeuniaux, 2008), we
think it is more likely that concepts are represented by sensory-motor activations and that these activations are reflected in their utterances.
Emotion. In addition to language processing, the role of sensory-motor activations for conceptual processing can also be clearly seen in how people
process emotion. The valence of a concept is the degree to which people have
positive or negative feelings about it. For example, people may have positive
feelings about kittens and negative feelings about snakes. Such feelings are
associated with approach and avoidance actions. If you like kittens, you have
a tendency to bring them closer to yourself, for example, by taking a step
in their direction or by picking one up and bringing it closer to your body.
Conversely, if you do not like snakes you have a tendency to increase the
distance between your body and the snake. In the laboratory, this is investigated by showing pictures or words with emotional valence and asking
participants to respond to the stimuli by moving a lever toward or away from
themselves, mimicking approach and avoidance actions respectively. People
find it easier to make approach responses when the stimulus has positive
valence than when it has negative valence and to make avoidance responses
when it has negative valence than when it has positive valence (Chen &
Bargh, 1999). The relation between valence and approach/avoidance actions
also works in the opposite direction. People tend to like meaningless stimuli
more after performing an approach action than after performing an avoidance reaction. Another important and related finding is mimicry. People have
a tendency to mimic other people’s facial expressions or bodily postures.
Researchers have suggested that mimicry underlies understanding of and
empathy with other people’s emotions (Niedenthal, Barsalou, Winkielman,
Krauth-Gruber, & Ric, 2005). People may use feedback from their own expression to recognize other people’s feelings. Mimicry is stronger between people who like each other than between people who do not like each other,
and when people are prevented from using mimicry their performance in
emotion recognition tasks drops. Thus, the close relation between emotion
understanding and emotional actions suggest that action is fundamental to
emotional concepts.

8

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

RECENT DEVELOPMENTS: THE GROUNDING OF ABSTRACT CONCEPTS
The grounded cognition approach has been criticized for dealing mainly
with the mental representations of concrete objects and actions. For concrete
objects and actions it is relatively easy to see how sensory-motor systems
might be involved in their representations. As we discussed above, there is
much evidence that conceptual processing and sensory-motor processing
interact, providing support for the idea that perception, action, and mental
representations share processing resources. In order to claim that cognition
is grounded in perception and action, however, we need to show that such
interactions exist not only for concrete objects such as apples but also for
more abstract concepts such as truth and problem solving.
Several ideas have been put forward to solve the grounding problem for
abstract concepts (Pecher, Boot, & Van Dantzig, 2011). The most thoroughly
investigated idea has its origin in cognitive linguistics, the Conceptual
Metaphor Theory (Lakoff & Johnson, 1980). On this account, people use
metaphors and image schemas to ground abstract concepts in sensory-motor
experiences. In language, people often use concrete domains to talk about
abstract concepts. For example, people talk about solving a problem in terms
of a journey that involves traveling from a starting point (the problem
situation) to a destination (the solution) along a path (the method that is
used to solve the problem). This can be seen in expressions such as to have
something get in one’s way or to reach the solution. People have very concrete
experiences with journeys, full of rich sensory-motor details. These concrete
experiences are mapped on the abstract situation of solving the problem,
giving it concrete structure. This account explains how abstract concepts
are grounded in sensory-motor systems because the mapping is based on
concrete physical experiences.
Starting with this linguistic evidence, cognitive scientists have further
investigated the psychological basis of Conceptual Metaphor Theory.
Although metaphors are observed in language, the theory claims that
metaphorical language is the expression of the underlying metaphorical
concepts. Thus, metaphors are not merely a linguistic phenomenon, but are
fundamental to conceptual processing. The theory distinguishes between
primary metaphors, or image schemas, and secondary metaphors. Especially
image schemas are relevant for grounded cognition. Image schemas are
basic patterns of embodied experience, such as up-down, inside-outside, and
balance. They are learned through multimodal experiences. The resulting
image schemas are analogue representations of embodied experiences in
space but they are not modality specific. Rather, they are more abstract
than direct sensory-motor experiences, and in this way can form a bridge
between direct physical interactions and abstract concepts.

Embodied Knowledge

9

Evidence for Conceptual Metaphor Theory is most convincing when image
schemas are activated in the absence of any linguistic reference to the image
schema. Studies have shown that mental representations of abstract concepts
result in activation of concrete image schemas. For example, researchers
have investigated whether the abstract concept power activates an up-down
image schema (Schubert, 2005). When people talk about power, they often
refer to powerful people as being higher than powerless people. Consistent
with this mapping, participants shifted their visual-spatial attention to
higher locations after reading a word referring to a person with much power
(e.g., president) than after reading a word referring to a person with little
power (e.g., servant). As a result, participants were better at identifying
an unrelated visual stimulus in congruent locations than in incongruent
locations (Zanolie et al., 2012). Similar findings have been reported on the
activation of spatial proximity for the concept similarity, up-down for quantity,
containment for category membership, spatial movement for time, and left-right
for valence (see Pecher et al., 2011, for an overview). All these studies provide
evidence for a role of image schemas for mental representations of abstract
concepts in contexts that contain no direct reference to the metaphorical
mapping.
Image schemas thus seem to play a role in de representation of abstract concepts. The question is to what extent Conceptual Metaphor Theory can fully
explain abstract concepts. Image schemas are very basic patterns, representing a single dimension with limited values. For example, the up-down image
schema represents only a concept’s vertical relative position. This alone is
not sufficient to represent the concept power. Rather, vertical position is an
embodied representation of relative differences in power between people. It
does not represent what it means to have power over someone, such as having the ability to make someone do something. A related issue is that there is
no one-to-one mapping between concepts and image schemas. For example,
many metaphors can be used for the concept love, such as journey, proximity,
and containment. Similarly, a single metaphor can be used for many concepts.
For example, the up-down metaphor can be used for power, quantity, valence,
and so on. This also illustrates that image schemas have only limited representational power. In other words, an image schema represents one property
of an abstract concept, and additional information is needed to represent the
full meaning of a concept.
Such additional details might be provided by situated concepts. In this
view, concepts develop from entire situations. A concept is a collection
of experiences that are combined into a representation. For example, the
concept power is based on many situated experiences that a person has with
power, and the representation of power likely occurs in a situated context.

10

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Abstract concepts often derive much of their meaning from an entire situation, more so than concrete concepts. Because situations have sensory-motor
properties, this proposal can also explain how abstract concepts might be
grounded. The representational power of such concept can be based entirely
on collections of experiences. This idea was recently further developed for
the emotion concepts fear and anger (Wilson-Mendenhall, Barrett, Simmons,
& Barsalou, 2011). Exemplar models of categorization (Hintzman, 1986;
Nosofsky, 1986) explain how individual episodes of experience are stored.
Abstraction occurs when a cue activates several experiences. These different
experiences are then combined into some sort of summary representation
for further processing. Thus, there is no need for the storage of abstract
representations, instead, abstractions are created as a response to a cue.
Another approach to grounding abstract concepts is to combine embodied
experiences with linguistic relations (Andrews, Vigliocco, & Vinson, 2009).
On this account, concepts are acquired through these two sources. Embodied
experiences provide sensory-motor grounding, but this may not be available
for abstract concepts. Knowledge is also acquired through language, however, and the availability of linguistic information is independent of how
concrete or abstract the concept is. They developed a model in which the
two sources are combined. In this model, abstract concepts are represented
as linguistic items that are related to other linguistic items and that, by inferences, are also related to sensory-motor information. Thus, abstract concepts
are grounded, albeit indirectly, in sensory-motor activations.
KEY ISSUES FOR FURTHER RESEARCH
Clearly, the question of how abstract concepts are grounded needs further
research. Although there is quite some evidence that image schemas are
important for abstract concepts, image schemas may be of limited value
for representations because they only represent certain properties. Richer
representations might consist of situational properties. Situations play a
role for all concepts, and perhaps even more so for abstract concepts than
for concrete concepts. In fact, it seems impossible to represent an abstract
concept such as power without the context. Thus, a theory of abstract
concepts should take situations into account. Situations should likely be
considered part of the concept rather than just background. For example,
experiments with problem solving tasks have shown that concrete details of
the situation are very important, even if they are irrelevant for the proper
solution of the problem. Research on transfer in problem solving tasks
such as in physics has shown that participants are more likely to use a
previously learned solution if problems share superficial properties, even
if the underlying structure is very different (Goldstone & Sakamoto, 2003).

Embodied Knowledge

11

One might argue that solving physics problems is a very difficult task for
novices. A question for future research is whether the same is true for more
automatic processes such as mental representation.
A second important issue is the role of language. First, language might
be important for the acquisition and perhaps also the representation of
abstract concepts. In Andrews et al.’s (2009) model concrete concepts are
grounded directly in sensory-motor experiences but are also represented by
the pattern of linguistic contexts in which they appear. Abstract concepts are
mostly represented by the linguistic contexts. One might wonder to what
extent representations can still be considered grounded in sensory-motor
processing if the grounding is indirect through other linguistic symbols.
Second, other researchers have even argued that linguistic data such as
word co-occurrences show effects of sensory-motor aspects of concepts. For
example, concepts that are perceptually similar occur in linguistically similar
contexts. Thus, linguistic relations might be sufficient for the representation
of meaning. One should take into account, however, that linguistic data
are produced by humans who already have rich concepts. Models of word
co-occurrences have as input texts such as internet news groups or textbooks
that are used in schools. All these texts are produced by people, and thus can
be expected to reflect the conceptual structure of the person producing the
test. Thus, rather than concluding that conceptual representation is based
on linguistic symbols, one could say they are expressions of underlying
concepts and have no bearing on the nature of those underlying concepts.
Attempts to use linguistic output to investigate the nature of the underlying
representations should therefore be regarded with some care.
A third issue is the need for formal models of grounded cognition. The
literature seems to be replete with claims that the grounded cognition
framework makes certain predictions but these claims are based on verbal
theories rather than detailed formal models. Formal models can give insight
into the feasibility of the approach. So far, few attempts have been made at
formal modeling. One important obstacle might be the operationalization of
sensory-motor input. If the input to a model is derived from verbal materials, such as feature production norms, one has to be aware that this is only
an approximation to actual sensory-motor representation. Computational
models of human cognition have played an important role in understanding
human behavior and may also help in better understanding how meaning is
represented.
CONCLUSION
To conclude, as many researchers from fields such as philosophy, linguistics,
and psychology have remarked, concepts need to be grounded. In the past

12

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

15 years, several ideas have been put forward on how concepts might be
grounded in sensory-motor processing. Inspired by these ideas, researchers
have obtained support for such ideas. Now, however, it seems we are reaching the boundaries. Some of these boundaries are mentioned above. If we
can jump these boundaries, the field will make further progress toward a
grounded view of cognition.
REFERENCES
Andrews, M., Vigliocco, G., & Vinson, D. (2009). Integrating experiential and distributional data to learn semantic representations. Psychological Review, 116, 463–498.
doi:10.1037/a0016261
Barsalou, L. W. (1993). Flexibility, structure, and linguistic vagary in concepts: Manifestations of a compositional system of perceptual symbols. In A. F. Collins, S. E.
Gathercole, M. A. Conway & P. E. Morris (Eds.), Theories of memory (pp. 29–101).
Hove, England: Erlbaum.
Barsalou, L. W. (1999). Perceptual symbol systems. Behavioral and Brain Sciences, 22,
577–660. doi:10.1017/S0140525X99002149
Bub, D. N., & Masson, M. E. J. (2010). Grasping beer mugs: On the dynamics of alignment effects induced by handled objects. Journal of Experimental Psychology: Human
Perception and Performance, 36, 341–358. doi:10.1037/a0017606
Chen, M., & Bargh, J. A. (1999). Consequences of automatic evaluation: Immediate
behavioral predispositions to approach or avoid the stimulus. Personality and Social
Psychology Bulletin, 25, 215–224.
Damasio, A. R. (1989). Time-locked multiregional retroactivation: A systems-level
proposal for the neural substrates of recall and recognition. Cognition, 33, 25–62.
doi:10.1016/0010-0277(89)90005-X
Glenberg, A. M. (1997). What memory is for. Behavioral and Brain Sciences, 20, 1–55.
doi:10.1017/S0140525X97000010
Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in action. Psychonomic
Bulletin & Review, 9, 558–565. doi:10.3758/BF03196313
Goldstone, R. L., & Sakamoto, Y. (2003). The transfer of abstract principles governing complex adaptive systems. Cognitive Psychology, 46, 414–466. doi:10.1016/
S0010-0285(02)00519-4
Harnad, S. (1990). The symbol grounding problem. Physica D, 42, 335–346.
doi:10.1016/0167-2789(90)90087-6
Hintzman, D. L. (1986). Schema abstraction in a multiple-trace memory model. Psychological Review, 93, 411–428. doi:10.1037//0033-295X.93.4.411
Klatzky, R. L., Pellegrino, J. W., McCloskey, B. P., & Doherty, S. (1989). Can you
squeeze a tomato? The role of motor representations in semantic sensibility
judgments. Journal of Memory and Language, 28, 56–77. doi:10.1016/0749-596X(89)
90028-4
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago, IL: Chicago University
Press.

Embodied Knowledge

13

Louwerse, M. M., & Jeuniaux, P. (2008). Language comprehension is both embodied
and symbolic. In M. De Vega, A. M. Glenberg & A. C. Graesser (Eds.), Embodiment and meaning: A debate (pp. 309–326). Oxford, England: Oxford University
Press.
Niedenthal, P. M., Barsalou, L. W., Winkielman, P., Krauth-Gruber, S., & Ric, F. (2005).
Embodiment in attitudes, social perception, and emotion. Personality and Social
Psychology Review, 9, 184–211. doi:10.1207/s15327957pspr0903_1
Nosofsky, R. M. (1986). Attention, similarity, and the identification-categorization
relationship. Journal of Experimental Psychology: General, 115, 39–57. doi:10.1037/
0096-3445.115.1.39
Patterson, K., Nestor, P. J., & Rogers, T. T. (2007). Where do you know what you
know? The representation of semantic knowledge in the human brain. Nature
Reviews: Neuroscience, 8, 976–987. doi:10.1038/nrn2277
Pecher, D., Boot, I., & Van Dantzig, S. (2011). Abstract concepts: Sensory-motor
grounding, metaphors, and beyond. In B. H. Ross (Ed.), The psychology of learning and motivation (pp. 217–248). Burlington, MA: Academic Press 10.1016/B9780-12-385527-5.00007-3
Pecher, D., Van Dantzig, S., Zwaan, R. A., & Zeelenberg, R. (2009). Language comprehenders retain implied shape and orientation of objects. Quarterly Journal of
Experimental Psychology, 62, 1108–1114. doi:10.1080/17470210802633255
Pecher, D., Zanolie, K., & Zeelenberg, R. (2007). Verifying visual properties in sentence verification facilitates picture recognition memory. Experimental Psychology,
54, 173–179. doi:10.1027/1618-3169.54.3.173
Pecher, D., Zeelenberg, R., & Barsalou, L. W. (2003). Verifying different-modality
properties for concepts produces switching costs. Psychological Science, 14, 119–124.
doi:10.1111/1467-9280.t01-1-01429
Pecher, D., Zeelenberg, R., & Raaijmakers, J. G. W. (1998). Does pizza prime coin?
Perceptual priming in lexical decision and pronunciation. Journal of Memory and
Language, 38, 401–418. doi:10.1006/jmla.1997.2557
Plaut, D. C. (2002). Graded modality specific specialisation in semantics: A computational account of optic aphasia. Cognitive Neuropsychology, 19, 603–639.
doi:10.1080/02643290244000112
Schubert, T. W. (2005). Your highness: Vertical positions as perceptual symbols of
power. Journal of Personality and Social Psychology, 89, 1–21. doi:10.1037/00223514.89.1.1
Simmons, W. K., Hamann, S. B., Harenski, C. L., Hu, X. P., & Barsalou, L. W. (2008).
fMRI evidence for word association and situated simulation in conceptual processing. Journal of Physiology, Paris, 102, 106–119. doi:10.1016/j.jphysparis.2008.03.014
Spence, C., Nicholls, M. E. R., & Driver, J. (2000). The cost of expecting events in the
wrong sensory modality. Perception & Psychophysics, 63, 330–336.
Stanfield, R. A., & Zwaan, R. A. (2001). The effect of implied orientation derived
from verbal context on picture recognition. Psychological Science, 12, 153–156.
doi:10.1111/1467-9280.00326
Van Dantzig, S., Pecher, D., Zeelenberg, R., & Barsalou, L. W. (2008). Perceptual processing affects conceptual processing. Cognitive Science, 32, 579–590. doi:10.1080/
03640210802035365

14

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Wilson-Mendenhall, C. D., Barrett, L. F., Simmons, W. K., & Barsalou, L. W. (2011).
Grounding emotion in situated conceptualization. Neuropsychologia, 49, 1105–1127.
doi:10.1016/j.neuropsychologia.2010.12.032
Wu, L., & Barsalou, L. W. (2009). Perceptual simulation in conceptual combination:
Evidence from property generation. Acta Psychologica, 132, 173–189. doi:10.1016/
j.actpsy.2009.02.002
Zanolie, K., Van Dantzig, S., Boot, I., Wijnen, J., Schubert, T. W., Giessner, S.
R., & Pecher, D. (2012). Mighty metaphors: Behavioral and ERP evidence that
power shifts attention on a vertical dimension. Brain and Cognition, 78, 50–58.
doi:10.1016/j.bandc.2011.10.006
Zeelenberg, R., Pecher, D., Shiffrin, R. M., & Raaijmakers, J. G. W. (2003). Semantic
context effects and priming in word association. Psychonomic Bulletin & Review, 10,
653–660.
Zwaan, R. A., Madden, C. J., Yaxley, R. H., & Aveyard, M. E. (2004). Moving
words: Dynamic representations in language comprehension. Cognitive Science, 28,
611–619. doi:10.1207/s15516709cog2804_5
Zwaan, R. A., Stanfield, R. A., & Yaxley, R. H. (2002). Language comprehenders mentally represent the shape of objects. Psychological Science, 13, 168–171. doi:10.1111/
1467-9280.00430

DIANE PECHER SHORT BIOGRAPHY
Diane Pecher is an Associate Professor at the Psychology Department of the
Erasmus University Rotterdam, the Netherlands. She holds a PhD from the
University of Amsterdam. Her research concerns conceptual memory and its
relation to language, perception, motor actions, and other types of memory.
RENÉ ZEELENBERG SHORT BIOGRAPHY
René Zeelenberg is an Associate Professor at the Psychology Department of
the Erasmus University Rotterdam, the Netherlands. He holds a PhD from
the University of Amsterdam. Rene spent several years at various US Universities as a Postdoc (Indiana University) and Visiting Scholar (Emory University, UCSD). His research concerns long-term memory, conceptual knowledge, and the influences of emotion on cognition.
Web sites:
www.memorylab.eu
eur.academia.edu/DianePecher
eur.academia.edu/ReneZeelenberg
www.brain-cognition.eu

Embodied Knowledge

15

RELATED ESSAYS
Theory of Mind and Behavior (Psychology), Amanda C. Brandone
Neighborhoods and Cognitive Development (Psychology), Jondou Chen and
Jeanne Brooks-Gunn
Language, Perspective, and Memory (Psychology), Rachel A. Ryskin et al.
Stereotype Content (Sociology), Beatrice H. Capestany and Lasana T. Harris
Culture and Cognition (Sociology), Karen A. Cerulo
The Inherence Heuristic: Generating Everyday Explanations (Psychology),
Andrei Cimpian
Cultural Differences in Emotions (Psychology), Jozefien De Leersnyder et al.
Micro-Cultures (Sociology), Gary Alan Fine
Economics and Culture (Economics), Gérard Roland
Language and Thought (Psychology), Susan Goldin-Meadow
Cultural Neuroscience: Connecting Culture, Brain, and Genes (Psychology),
Shinobu Kitayama and Sarah Huff
Resource Limitations in Visual Cognition (Psychology), Brandon M. Liverence and Steven L. Franconeri
Immigrant Sociocultural Adaptation, Identification, and Belonging (Sociology), Sarah J. Mahler
Concepts and Semantic Memory (Psychology), Barbara C. Malt
Emerging Trends in Culture and Concepts (Psychology), Bethany Ojalehto
and Douglas Medin
Neural and Cognitive Plasticity (Psychology), Eduardo Mercado III
A Bio-Social-Cultural Approach to Early Cognitive Development: Entering
the Community of Minds (Psychology), Katherine Nelson
Social Classification (Sociology), Elizabeth G. Pontikes
Attention and Perception (Psychology), Ronald A. Rensink
Class, Cognition, and Face-to-Face Interaction (Sociology), Lauren A. Rivera
Speech Perception (Psychology), Athena Vouloumanos
Theory of Mind (Psychology), Henry Wellman