Neuron
e e
U
ity
al
cumscribed brain system associated
with social cognition: the social brain.
research in social neuroscience came
gether this exciting work. There are three
key issues to be discussed. First, there
institutions. Second, there is the problem
ously taught with little if any connection
to our biological nature. Now we can con-
a genetic basis, it is plainly not sufficientlatti et al., 1996). This observation reveals
that there is a mechanism within the social
brain with the potential to enable learning
exhaustive review by Ebstein et al.
(2010), many unique aspects of human
social behavior are, at least partially,
linked to basic mechanisms, what con-
trols them and what doesn’t, how to
explain their origin in development, andthrough imitation and to infer intentions
from action observation. Researchers
under genetic control, and here remark-
able work is being done at the molecular
how to account for their elaboration
through culture.from the discovery of mirror neurons in
the monkey. These neurons fire both
when the monkey performs a specific
goal-directed action and also when the
monkey sees the experimenter perform-
ing the same goal-directed action (Rizzo-
of the distinction between social and
nonsocial cognition. Third, there is the
problem of determining what is special,
if anything, about human social cognition.
Let us start with the gap between mole-
cules and behavior. As is revealed in the
to explain social behavior. As Sokolowski
(2010) points out, genes don’t determine
behavior; they encode molecules that
build brain tissue. When we try to under-
stand empathy or political attitudes, we
need to know how these concepts areThe mentalizing deficit hypothesis en-
couraged the search for underlying neural
mechanisms and posed the question of
whether this was unique to the human
brain.
A second impetus for the increase in
is the problem of how to bridge the gap
between genes and neurons on one side
and social behavior on the other, which
in the case of humans includes a vast
array of historically enduring accomplish-
ments, including the existence of cultural
template a genetic predisposition that
biases you to vote for one political party
rather than another and a predisposition
for you to make more or less altruistic
decisions. However, exciting as the idea
is that proclivities in social behavior haveOverview
Learning from Oth
Introduction to th
on Social Neurosc
Chris Frith
1,3,
* and Uta Frith
2,3
1
Wellcome Trust Centre for Neuroimaging at
2
Institute of Cognitive Neuroscience, Univers
3
Interacting Minds Group, Center of Function
*Correspondence: c.frith@ucl.ac.uk
DOI 10.1016/j.neuron.2010.03.015
The last decade has seen a dramatic rise
of interest in the study of social neurosci-
ence. Two observations have had a major
role in driving this interest. First, there was
the discovery that autism is associated
with specific difficulties in social cogni-
tion, while nonsocial cognition, and in
particular IQ, can remain intact (Frith,
1989; Hermelin and O’Connor, 1970).
This discovery was made during the time
when the information-processing revolu-
tion was transforming the behavioral
sciences and when researchers were
striving to find mechanisms underlying
behavior, a significant departure from a
preoccupation with surface appearances.
This change of approach led to the pro-
posal of a mechanism that could explain
some of the characteristic social impair-
ments of autism, a lack of Theory of
Mind, or inability to mentalize (Baron-
Cohen et al., 1985). The case of autism
lent weight to the idea that there is a cir-ers:
Special Review S
ience
niversity College London
College London
ly Integrative Neuroscience, Aarhus University
were thus encouraged to try and specify
such mechanisms (e.g., Kilner et al.,
2007) in animals, including humans, where
learning from conspecifics by imitation,
emulation, or mimicry is pervasive.
The contributors to this special issue on
social neuroscience review the research
of the last decade and reveal how very
much more there is to social cognition
and that these earlier discoveries were
only the beginnings of a vast enterprise.
This enterprise, at least at first glance,
focuses on learning from others. This
kind of social behavior can be observed
in animals from fruit flies to humans. The
mechanisms underlying this behavior are
beginning to be revealed at the molecular
level. In addition to the wide range of
animal species considered in this special
review issue, a wide range of approaches
and interpretations are applied to the
observations. In this overview we will
present our own ideas for drawing to-Neuron 6ries
level. However, this work is far removed
from explaining, say, cultural learning. It
is even far removed from explaining the
processes by which genetic factors exert
control on individual differences in social
behavior. Twin studies show that there
are complex interactions with nongenetic
influences, and the study of epigenetic
effects is thriving. One example of an
epigenetic effect on social behavior is
seen in honey bees: When worker bees
feed larvae with royal jelly, the expression
of genes involved in growth and metabo-
lism is changed, and this leads to the
development of new queens (Sokolowski,
2010).
Examples of behavior under partial
genetic control mentioned by Ebstein
et al. (2010) and Insel (2010) include
economic decision making and political
attitudes. Such behaviors were previously
considered as typically human achieve-
ments, culturally prescribed and precari-5, March 25, 2010 ª2010 Elsevier Inc. 739

io-
e
so
ng
er
p-
an
m
a-
ge
he
l.,
in
ed
ral
tive mechanism, even if not explicitly clas-
sified as such by the author. Insel (2010)
describes some of the links between
genes and brain systems in his discussion
of the role of oxytocin, but recognizes
the problem of‘‘the great dark matter of
social neuroscience’’that lies between
perception and action. Adolphs (2010)
also recognizes this problem and
suggests that‘‘analysis at the level of the
brain could serve as a unifying base.’’
However, this aim will be difficult to fulfill
in light of the idea that there are homologs
of social capacities in species with very
different brains. Animals, such as birds,
cetaceans, and mammals, possessing
very different brain structures, can never-
exa
emp
ces
and
uno
at t
that
ces
som
sam
they
be r
Suc
com
Lea
Lea
e
n
v
i
r
o
n
m
e
n
t
t
ig t
e
e
Neuronmunication (Frith et al., 1991).
The critical question is what do the b
logical roots allow the mind to do? W
imagine there is some parsimony,
that we allow for a many-to-one mappi
from biology to cognition. On the oth
hand, we allow for a one-to-many ma
ping from cognition to behavior. We c
assume that a single mental mechanis
(often in interaction with other mech
nisms) can be responsible for a lar
variety of behaviors. For example, t
ability to learn language (Fitch et a
2010) has different manifestations
different species and can be assess
with different tests at both the neu
and behavioral level.The Mediating Role of the
Cognitive Level of Description
We were fascinated to learn from this
special review issue of the extent to
which the same social processes
can be observed in so many different
species. This tells us that, at some
level, which we call the cognitive level,
evolution has led to homologous solu-
tions dealing with the problems and
advantages that arise from living with
conspecifics.
Figures 1 and 2 are simple illustra-
tions of how to conceptualize a cog-
nitive level of description and how it
fits with the whole enterprise of social
neuroscience (diagrams adapted
from Morton and Frith, 1995).
We see in these figures that cognition is
situated in between brain and behavior,
and by virtue of this position forms a link
between these two rather distinct levels
of explanation. This is important because
there is no one-to-one mapping when
crossing levels. For example, as Figure 2
shows, at the biological level we can allow
for a variety of genetic pathways and
mechanisms. These pathways lead to
the development and maintenance of dis-
tinct parts of the central nervous system.
Different mechanisms can be modeled,
as here in terms of multiple neural net-
works underlying the cognitive process
termed mentalizing, which can be as-
sessed by such diverse behaviors as joint
attention, deception, and ostensive com-
F
L
NWhile vocalization appears to be crucial
for the development of language and
communication, we are still only guessing
what other neurocognitive mechanisms
are being enabled or changed by evolu-
740 Neuron 65, March 25, 2010 ª2010 Elsevtionary pressures on social behavior.
Mentalizing and the corepresentation of
action and observation are two very
recent examples of such mechanisms,
which had hardly been envisaged earlier.
The search for neural systems underlying
these mechanisms has been remarkably
successful, while the search for the
genetic foundations for the origin of our
ability to learn language has been nothing
less than a triumph (Fitch et al., 2010).
For now, however, we need to accept
that there are huge gaps between brain
and mind and behavior in almost any of
the social behaviors and almost any of
the biological mechanisms discussed
by the present reviewers.
In the papers in this issue, there are a
number of concepts that describe a cogni-
brain
genes
behav
cogni
ure 1. A Simple Conceptualization of Differen
vels of Explanation in Comparative Social
urosciencetheless all be said to have some under-
standing of the point of view of others,
including possibly their mental states.
Furthermore, evidence of learning from
the observation of conspecifics can be
defi
teriz
can
in w
soc
ier Inc.mple, Insel (2010) talks of the ‘‘brain
loying specific receptors for the pro-
sing of social information.’’ Byrne
Bates (2010) talk of ‘‘representing
bservable causal factors.’’ Analysis
he cognitive level permits recognition
the same sensory signals can be pro-
sed in different ways. For example, if
e species are not able to form the
e mental representations, then what
do with the same information may
adically different (Byrne et al., 2004).
h recognition is critical when we start
paring different species.
rning by Observation
rning from others is the most basicfound from fruit flies to humans. There
are, however, considerable variations
in the sophistication of this learning.
The cognitive level makes it simple
to accommodate the possibility that
a variety of neural mechanisms can
all underpin critical processes that
allow learning from others. At this
level, mental processes, whether in-
stantiated in the bird brain, the
monkey brain, or the human brain,
can be understood to serve the
same aims. For example, all these
creatures send out signals of commu-
nication and modulate their behavior
in the presence of conspecifics. We
believe that the cognitive level can
be developed to fill the gap between basic
molecular processes and intuitively per-
ceived social abilities, since the vocabu-
lary of cognition, and in particular the
computational models associated with
this level, can be applied equally to neural
as to mental processes.
We note that we are using the term
cognitive in its modern sense. We are
not using cognition in the restricted sense
of knowledge as opposed to emotion or
will. We are certainly not using cognition
in the sense of conscious processes. We
are using it as in‘‘cognitive neurosci-
ence’’; a mechanistic account of neural
and psychological processes using terms
derived from computational theory. Such
terminology can be found in many of the
contributions to this review issue. For
iour
ion
Overviewnition of social cognition that charac-
es all social species. However, we
learn from others and about others
ays that do not qualify as implicating
ial processes at all. Very clear