Magazine
R593
What and how do they eat? Most
mode of processing, the function of
Correspondence
species are thought to be carnivorous
which is to extract parameters that
grazers, feeding on sponges,
are relevant to action, and to generate
cnidaria or even mol uscs. Abyssal
the corresponding motor commands”
(Mis)understanding
pycnogonids have been observed in
(p. 320), as opposed to “‘semantic’
association with anemones at whale-
analysis [which is] performed in the
mirror neurons
and wood-fal sites on the sea bed
temporal lobe” (p.314) [3]. Thus, the
(Figure 2). They can both walk and
meaning of objects is not coded in F5,
swim, although walking is extremely
Gregory Hickok1 and Marc Hauser2
although clearly, “the semantic system
slow. Some species ‘swim’ by tucking
can influence the pragmatic system”
up their legs and al owing themselves
It is hard to imagine a class of
(p. 320) [3] (for example, we want to
to sink rapidly, others beat their legs.
neurons that has generated more
reach for food not snakes).
Tidal species can sense changes in
excitement than mirror neurons,
Mirror neurons were discovered
water pressure and alternate between
cells discovered by Rizzolatti and
within this same circuit and found to
swimming and walking accordingly.
colleagues [1] in macaque area F5
have similar sensorimotor properties
Because they do not have a planktonic
that fire both during action execution
[1,4]. It was even suggested that
larval phase, they do not disperse far,
and action observation. We suggest,
“the actions performed by other
although at least one shal ow-water
however, that the interpretation
monkeys must be a very important
species may show seasonal migration
of mirror neurons as supporting
factor in determining action selection”
into deeper water.
action understanding was a wrong
(p. 179) [4] and that “the [motor]
turn at the start, and that a more
vocabulary of F5 can be addressed in
When did they first appear? There
appropriate interpretation was lying
two ways: by objects and by events
are very few pycnogonids in the fossil
in wait with respect to sensorimotor
[actions]” (p. 317) [3]. Thus, the
record. The earliest apparent sea spider learning. We make a number of
theoretical and empirical pieces were
is a larva found among the exquisitely
arguments, as follows. Given their
in place to interpret mirror neurons
preserved Upper Cambrian “Orsten”
previous work, it would have been
as sensorimotor association cells
deposits from around 500 mil ion years
natural for Rizzolatti’s group to
relevant to action selection, just like
ago. A later, adult sea spider has been
interpret mirror neurons as involved
object-oriented cells (Figure 1).
found in the Silurian (425 mil ion years
in action selection rather than action
But this interpretation was not
ago), and others have been reported in
understanding. They did not make this
considered — why?
the Devonian and the Jurassic. Some of assumption because, at the time, the
It was the mirroring property
the Devonian specimens are relatively
data suggested that monkey behavior
of mirror neurons that steered
primitive and retain a tail.
did not support such an interpretation. investigators away from a
Recent evidence shows that monkeys
straightforward sensorimotor
What is the future? Final resolution of
do, in fact, exhibit behaviors that
interpretation. The logic was, if mirror
their phylogenetic position wil require
support this alternative interpretation.
actions (for example, imitation) are not
substantial amounts of morphological
Thus, the original basis for claiming
in the species’ repertoire, then mirror
and DNA data from a wide range of
that mirror neurons mediate
neurons can have no motor selection
species. This wil not only resolve
action understanding is no longer
function. Rizzolatti and Craighero
one of the thorniest questions in
compelling. There are independent
used this argument, pitting “two main
evolutionary biology, it wil also open
arguments against the action
hypotheses” of mirror neuron function,
the door to functional studies of these
understanding claim and in support
imitation and action understanding;
enigmatic and bizarre animals.
of a sensorimotor learning origin
because macaques do not imitate,
for mirror neurons. Therefore, the
they argued, mirror neurons must
Where can I find out more?
action understanding theory of mirror
support action understanding
Arango, C.P., and Wheeler, W.C. (2007). Phylogeny
neuron function requires serious
(p. 172) [1]. However, these authors,
of the sea spiders (Arthropoda, Pycnogonida)
based on direct optimization of six loci and
reconsideration, if not abandonment.
and the field generally, have failed
morphology. Cladistics 23, 255–293.
Mirror neurons were discovered
to notice that other forms of mirror
Arnaud, F., and Bamber, R.N. (1987) The biology of
in the context of research aimed
actions are in the macaque motor
Pycnogonida. Adv. Marine Biol. 24, 1–96.
Bamber, R. (2007). A holistic re-interpretation of
at understanding how the visual
repertoire. For example, field studies
the phylogeny of the Pycnogonida Latreille,
properties of objects are integrated
show that rhesus monkeys perceive
1810 (Arthropoda), in Z.-Q. Zhang and W.A.
Shear (eds). Linnaeus Tercentenary: Progress in
with motor codes for action. Cells
human gestures as goal-directed,
Invertebrate Taxonomy. Zootaxa 1688, 295–312.
in area F5 were found to respond to
including those that mimic the rhesus
Brenneis, G., Ungerer, P., and Scholtz, G. (2008).
visually presented objects as well
monkeys’ species-specific signal for
The chelifores of sea spiders (Arthropoda,
Pycnogonida) are the appendages of the
as during grasping actions towards
coalition recruitment [5]. Macaques
deutocerebral segment. Evol. Dev. 10, 717–724.
those objects. The interpretation
also engage in contagious yawning,
Dunlop, J.A., and Arango, C.P. (2004). Pycnogonid
affinities: a review. J. Zool. Syst. Evol. Res. 43,
of this circuit was that it coded a
where perception of another’s yawn
8–12.
“vocabulary of motor acts and that
triggers a yawn in the observer [6].
http://www.marinespecies.org/pycnobase/
this vocabulary can be accessed by …
Further, experimental work has found
King, P.E. (1973). Pycnogonids. Hutchinson.
visual stimuli” (p. 491) [2] and that it
that another’s grasping actions toward
was critical for “learning associations,
one of two food receptacles serves
Faculty of Life Sciences, University
of Manchester, Oxford Road,
including arbitrary associations
as a cue to goal-directed grasping
Manchester M13 9PT, UK.
between stimuli and [motor] schemas” toward that same receptacle [7] — an
E-mail: cobb@manchester.ac.uk
(p. 317) [3]. This is a “‘pragmatic’
experimental situation reminiscent

---

Current Biology Vol 20 No 14
R594
A
Motor output
In summary, a sensorimotor theory
can explain the response properties
of mirror neurons, does so more
straightforwardly, and does not
DORSAL STREAM
suffer the empirical roadblocks of the
action understanding theory [5,10].
Object-oriented sensorimotor integration
It is time to reconsider mirror neuron
Action-oriented sensorimotor integration
function and the neural basis of action
Action “understanding”
understanding.
Visual
feature
References
input
1. Rizzolatti, G., and Craighero, L. (2004). The
mirror-neuron system. Annu. Rev. Neurosci. 27,
Semantic input
(objects only)
VENTRAL STREAM
169–192.
2. Rizzolatti, G., Camarda, R., Fogassi, L.,
Gentilucci, M., Luppino, G., and Matel i, M.
Object “understanding”
(1988). Functional organization of inferior area
6 in the macaque monkey. II. Area F5 and the
control of distal movements. Exp. Brain Res. 71,
491–507.
3. Jeannerod, M., Arbib, M.A., Rizzolatti, G., and
Sakata, H. (1995). Grasping objects: the cortical
mechanisms of visuomotor transformation.
B
Motor output
Trends Neurosci. 18, 314–320.
4. di Pel egrino, G., Fadiga, L., Fogassi, L., Gal ese,
V., and Rizzolatti, G. (1992). Understanding
motor events: a neurophysiological study. Exp.
Brain Res. 91, 176–180.
DORSAL STREAM
5. Hauser, M., and Wood, J. (2010). Evolving the
capacity to understand actions, intentions, and
goals. Annu. Rev. Psychol. 61, 303–324, C301.
Object-oriented sensorimotor integration
6. Paukner, A., and Anderson, J.R. (2006). Video-
induced yawning in stumptail macaques
Action-oriented sensorimotor integration
(Macaca arctoides). Biol. Lett. 2, 36–38.
7. Wood, J.N., Glynn, D.D., Phil ips, B.C., and
Hauser, M.D. (2007). The perception of rational,
Visual
goal-directed action in nonhuman primates.
feature
Science 317, 1402–1405.
input
8. Range, F., Viranyi, Z., and Huber, L. (2007).
Selective imitation in domestic dogs. Curr. Biol.
Semantic input
VENTRAL STREAM
17, 868–872.
9. Heyes, C. (2010). Where do mirror neurons
come from? Neurosci. Biobehav. Rev. 34,
Object “understanding”
575–583.
10. Hickok, G. (2009). Eight problems for the mirror
Action “understanding”
neuron theory of action understanding in
monkeys and humans. J. Cogn. Neurosci. 21,
Current Biology
1229–1243.
1Center for Cognitive Neuroscience,
Figure 1. Schematic models of dorsal and ventral stream function.
Department of Cognitive Sciences,
(A) The current dominant model [1], which holds that object- and action-oriented processes
University of California, Irvine, CA 92617,
for sensorimotor integration and ‘understanding’ are organized differentially, with action un-
USA. 2Departments of Psychology and
derstanding part of the dorsal sensorimotor stream and object ‘understanding’ part of the
Human Evolutionary Biology, Harvard
ventral stream. (B) A more conventional model in which object- and action-oriented processes
University, 33 Kirkland Street, Cambridge, MA
for sensory-motor integration and understanding are organized similarly. Both models assume
02138, USA.
that semantic information from the ventral stream can modulate sensorimotor processes in the
E-mail: greg.hickok@gmail.com
dorsal stream.
of the mirror neuron studies. Even
the motivating argument for the
The editors of Current Biology
domesticated dogs mirror goal-
action understanding theory over a
directed actions of a model dog
sensorimotor theory (for example [9])
welcome correspondence on
[8]; one would expect to find mirror
does not hold.
any article in the journal, but
neurons in dogs given this behavioral
Can we distinguish the sensorimotor
reserve the right to reduce
evidence. And lastly, rhesus monkeys
and action understanding theories
comprehend actions that they are
of mirror neurons? Yes: empirical
the length of any letter to be
physically incapable of producing. In
findings favor the sensorimotor

published. Al Correspondence
particular, though rhesus monkeys
account by showing that action

containing data or scientific
do not throw, they can recognize a
understanding and motor system
throwing action in humans, realizing
function dissociate [10], that motor

argument wil be refereed.
that throwing a rock is dangerous
actions alone are insufficient to

Queries about articles for
whereas throwing food is not [5].
explain action understanding [5], that

consideration in this format
Observed actions can serve as
animals comprehend many actions
important inputs to action selection,
that they cannot execute [10], and that
should be sent by e-mail to
including, but not necessarily
sensorimotor learning can transform
cbiol@current-biology.com
limited to, mirror actions. Therefore,
the mirror system [9].

---

Document Outline

• (Mis)understandingmirror neurons
  • References

---