MNS and ToM.pdf

BRAIN RESEARCH REVIEWS 54 (2007) 286 – 293

available at www.sciencedirect.com

www.elsevier.com/locate/brainresrev

Review

The human mirror system: A motor resonance theory of

mind-reading

Zarinah K. Agnew ⁎ , Kishore K. Bhakoo, Basant K. Puri

MRC Clinical Sciences Center, Imperial College London, UK

ARTICLE INFO

ABSTRACT

Article history:

Accepted 18 April 2007

Available online 24 April 2007

Electrophysiological data confirm the existence of neurons that respond to both motor and

sensory events in the macaque brain. These mirror neurons respond to execution and

observation of goal-orientated actions. It has been suggested that they comprise a neural

basis for encoding an internal representation of action. In this paper the evidence for a

parallel system in humans is reviewed and the implications for human theory of mind

processing are discussed. Different components of theory of mind are discussed; the

evidence for mirror activity within subtypes is addressed. While there is substantial

evidence for a humanmirror system, there are weaknesses in the attempts to localize such a

system in the brain. Preliminary evidence indicates that mirror neurons may be involved in

theory of mind; however, these data by their very nature are reliant on the presence, and

precise characterization, of the human mirror system.

Keywords:

Cognitive neuroscience

Functional magnetic

resonance imaging

Mirror neuron

Mirror system

Motor intention

Theory of mind

Contents

1. How can cognition emerge from neurons? .........................................287

2. Human mirror activity ....................................................287

2.1. Electrophysiological studies

—

where is the human mirror system? .....................287

3. How I know why you do what you do ............................................288

3.1. Brain areas involved in ToM and mirror function..................................288

3.2. Targeting areas involved in understanding action intention . . ..........................290

3.3. Mirror activity in people with absent or attenuated ToM..............................290

4. The role of mirror neurons in different types of theory of mind . . . ..........................290

5. Conclusions ..........................................................291

References..............................................................291

—

Corresponding author. Medical Research Council Clinical Science Centre, Imperial College London, Robert Steiner MR Unit, Hammersmith

Hospital, Du Cane Road, London W12 0NN, UK.

E-mail address: z.agnew@csc.mrc.ac.uk (Z.K. Agnew).

Abbreviations: ASD, Autistic spectrum disorder; EEG, electroencephalogram; fMRI, functional magnetic resonance imaging; MEG,

magnetoencephalogram; MEP, motor evoked potential; MN, mirror neuron; PET, positron emission tomography; PMC, premotor cortex;

STS, superior temporal sulcus; ToM, theory of mind

doi: 10.1016/j.brainresrev.2007.04.003

–

dohumanshaveamirrorsystem?.........................287

2.2. Functional MRI and PET studies

⁎

0165-0173/$

BRAIN RESEARCH REVIEWS 54 (2007) 286 – 293

287

1. How can cognition emerge from neurons?

of actions produced motor evoked potentials (MEPs) in the

muscles involved in that movement. This has since been

replicated ( Strafella and Paus, 2000 ), and a temporal correlation

between the MEPs recorded and the action observed has also

been reported ( Gangitano et al., 2001 ). Stimulation of the

median nerves during manipulation of an object results in

suppression of a post-stimulus rebound effect when recording

electrical oscillations from precentral motor cortex ( Salmelin

andHari, 1994 ) usingMEG. This post-stimulus rebound event is

suppressed to a lesser extent in response to action observation

alone ( Hari et al., 1998 ). These results are interpreted as

evidence for primary motor cortex activation in response to

action execution and observation.

Together these studies provide strong evidence for a human

mirror system in the central nervous system which appears to

originate from the motor system. The function that this system

encodes is the subject ofmuch discussion andwill be addressed

in later sections. Inorder to assess further the specific structures

of the brain that are involved in this system, a different range of

techniques with higher spatial acuity is required.

provide a link

between abstract concepts and single neuron activity ( Quiroga

et al., 2005 ), and the demonstration of plasticity at the synaptic

level provided a mechanism by which information can be

encoded across time. One such advance is the discovery of

mirror neurons (MNs) in the premotor cortex of the macaque.

This finding and its application to understanding information

processing in the human brain will be the focus of this essay.

Mirror neurons respond to the execution of an action, and

to the observation of a conspecific carrying out that same

action ( di Pellegrino et al., 1992; Gallese et al., 1996; Rizzolatti

and Craighero, 2004; Rizzolatti et al., 1996a,b ).

First identified in the brain of macaque monkeys, these cells

have stimulated a wave of research by authors attempting to

identify the equivalent cells in the human brain. This section

will review critically the evidence for mirror neuron activity in

humans and discuss the potential role of mirror neurons in

human behavior. Thiswill include a full discussion of the role of

mirror neurons in theory of mind (for a more detailed summary

of mirror neurons in other aspects of brain function such as

language and imitation refer to Rizzolatti and Craighero, 2004 ).

Grandmother cells

’

2.2. Functional MRI and PET studies

—

where is the

human mirror system?

The evidence suggests that the human mirror system stems

from activity in the inferior parietal lobe, inferior frontal gyrus

(including Broca's area) and superior temporal sulcus (STS)

( Rizzolatti and Craighero, 2004 ). The majority of these studies

employ functional magnetic resonance imaging (fMRI). This

technology provides information about regional cerebral blood

flow in the brain in response to a range of stimuli (blood

oxygen level dependent or BOLD effect). From this information

it is possible to localize neuronal activity to regions of the

brain; however, the neural basis of the BOLD response is far

from clear ( Logothetis, 2003 ). Interestingly, the human mirror

system appears to overlap considerably with those from non-

human primate data; the mirror circuit proposed in the

macaque involves projections from the superior temporal

sulcus ( Jellema et al., 2000; Perrett et al., 1990 ) to inferior

parietal lobe (PF) ( Gallese et al., 2002 ) and ventral premotor

cortex (V5) ( di Pellegrino et al., 1992; Rizzolatti et al., 1996a,b ).

One of the earliest fMRI studies compared action observa-

tion, imitation and execution. They found that action execu-

tion and observation resulted in activity in left frontal

operculum (Broca's area) and right anterior parietal area.

Imitation resulted in additional activity in parietal operculum

( Iacoboni et al., 1999 ). The authors report increased activity

when the action was imitated which implies that the mirror

system may be involved in imitation (see Section 3.1). The

protocol used in this study involved a simple finger move-

ment. There is controversy as to the specific stimuli that will

target the human system. Mirror activity is only reported in

response to object- and goal-related actions and not mean-

ingless intransitive movements in the macaque ( Rizzolatti

and Craighero, 2004 ). These data subsequently may not

actually reflect what we might call

2. Human mirror activity

2.1. Electrophysiological studies

—

do humans have amirror

system?

Sensory information received during observation of actions is

encoded in terms of a motor echo. This has been demonstrat-

ed by electroencephalography (EEG), magnetoencephalogra-

phy (MEG) and single-cell recordings from human brains. The

only study to demonstrate mirror activity at a neuronal level

comes in the formof an investigation into nocioception: single

cell recordings from the anterior cingulate of patients undergo-

ing surgery demonstrate activity in response to both percep-

tion and observation of pain ( Hutchison et al., 1999 ). Given that

opportunities to carry out such studies are rare, EEG provides a

more convenient alternative to invasive techniques.

Normal mu wave activity undergoes dysynchronization in

response to both action execution and observation ( Cohen-

Seat et al., 1954; Gastaut and Bert, 1954 ) as is demonstrated by

placing electrodes placed on the scalp. A more sensitive form

of EEG has since confirmed these findings; Cochin et al. report

that observation and execution of finger movements resulted

in reduced power of alpha wave activity (7.5

. Cortical

excitability studies indicate that the MN system is sensitive to

all the movements that form an action, not just the action as a

whole ( Rizzolatti and Craighero, 2004 ). However, it is not

possible to conclude on this in the absence of intracellular

mirror activity

’

10.5 Hz) in 20

subjects. They localized this decrease to motor and frontal

cortices ( Cochin et al., 1998, 1999 ). Furthermore, Fadiga and

colleagues ( Fadiga et al., 1995 ) demonstrated that observation

–

A major aim of cognitive neuroscience is to explain how the

brain functions in terms of its cellular building blocks, namely

neurons. In order to understand how the human experience

emerges fromthe neuronal structure of the brain, wemust link

findings fromcellular and cognitive neuroscience. A successful

example of this approach can be seen in the multidisciplinary

study of hierarchical visual processing. Periodically certain

developments in neuroscience research allow us to grasp this

problem in a novel way;

‘

288

BRAIN RESEARCH REVIEWS 54 (2007) 286 – 293

recordings in humans. These properties are distinct from

those seen in non-human primates and such differences may

explain why humans, but not macaques, have abilities such as

theory of mind or advanced imitation.

Grezes et al. (2003) have also compared action observation

and execution using fMRI. They found that object-related

action observation and execution resulted in increased

activation in bilateral dorsal premotor cortex (PMC), intrapar-

ietal sulcus, superior temporal sulcus (STS) and right parietal

operculum (SII). This study employed the use of an abstract

object for the participants to grasp, which, while taking into

account the object-related nature of mirror neurons, may not

quite fulfil the criteria for goal-related. Furthermore, Gallese

et al. (2004) have proposed that the internal object represen-

tation may be influenced by one's experience or interaction

with it. The use of an abstract

of the rest of this essay and is discussed in reference to

recognition of intention.

3. How I know why you do what you do

which has no

clear associated action may influence the activity seen in the

brain. In addition to this, televized stimuli have been shown to

produce a diminished response in the macaque model ( Ferrari

et al., 2003; Keysers and Perrett, 2004 ) in certain brain regions,

which this study also used. Thus this paradigmmay not have

targeted putative mirror neurons in an ideal manner. A final

and well-structured study investigated the role of context on

mirror activity by exposing participants to videos of two types

of mug grasping actions ( Iacoboni et al., 2005 ). Unfortunately

however, they did not investigate action execution in this

study thus few conclusions can be made about mirror activity.

A positron emission tomography (PET) study reported that

action recognition was associated with increased activation in

the left superior temporal sulcus and caudal inferior frontal

gyrus ( Rizzolatti et al., 1996a,b ). They did not however contrast

grasp observation with grasp execution, which is the crucial

comparison that would highlight mirror activity. They con-

cluded that human action recognition is represented by a

pattern of activity in middle temporal gyrus, STS and left

inferior frontal gyrus including Broca's area. However, at no

point during their discussion or conclusions do the authors

discuss the relevance of their results in terms of humanmirror

activity. Of interest was the activation seen in Broca's area in

response to grasping observation, as this is an area previously

thought to be dedicated to language.

A final but important point to note is that the spatial

resolution of fMRI, which is higher than that achieved

through PET, is in the order of millimetres. As a result it

cannot be concluded that an increase in the BOLD response to

action execution and observation is conclusive evidence for

mirror neurons in the human brain. On the basis of current

electrophysiological and imaging data, we are not able to

speculate on individual neurons. Accordingly, we are limited

to discussing brain areas, as pointed out by Arbib (2005, http://

www.interdisciplines.org/mirror/papers/4/2 ). Whilst it is hard

to see why mirror neurons would exist in the macaque brain

and not the human brain, ultimately there is no direct

evidence of human mirror neurons that respond to action.

In summary, there appears to be substantial evidence for a

mirror system within the central and peripheral nervous

system. However, the evidence attempting to localize this

mirror network has some weaknesses; further confirmation

and characterization is required. The next question that begs

to be addressed is that of function. This comprises the focus

manipulandum

’

It has been suggested that mirror neurons or the human

equivalent may be involved in understanding the intentions of

others ( Gallese and Goldman, 1998 ). One can instinctively see

how this might occur if components of one's own motor

system echo an observed action. In the world of neuropsy-

chology, this ability, unique to humans in its more developed

form, is known as theory of mind (ToM).

Theory of mind refers to two concepts: the knowledge that

other animals have mental states which may differ from our

own; and the ability to infer what these internal states may be.

Such states refer to beliefs, goals, intentions or emotions. This

term covers a range of skills which may or may not have

overlapping neural bases. The term theory of mind fails to

distinguish between these two concepts and subsequently

tends to imply that possession of a

or of other

minds in itself constitutes the ability to infer the states of

other minds. This is doubtfully the case; the knowledge that

other people or animals have internal states that differ from

our own is more likely a prerequisite for the discrete ability to

make correct inferences based on this knowledge. The

neuronal mechanisms by which we are able to manipulate

this knowledge are also likely to be distinct. In our case, this

manipulation may be one of motor simulation as it has been

suggested that mirror neurons may be involved in theory of

mind ( Gallese and Goldman, 1998 ) and there is accumulating

evidence to substantiate this hypothesis.

There are at least three prominent ways in which the role of

mirror neurons in ToM processing can be assessed. Firstly it is

possible to compare brain areas involved in both ToM and mirror

activity. If ToM relied onmirror neuron activity, areas involved in

ToM might be expected to display mirror activity. Secondly,

studies could be designed to specifically tap into the role ofmirror

neurons inunderstanding the intention behind the action. A final

approach which would support this theory, would be to look at

mirror neuron activity in people with absent or abnormal theory

of mind capabilities. The evidence supporting these lines of

investigation will be assessed in the following sections.

theory of mind

’

3.1. Brain areas involved in ToM and mirror function

A network including medial prefrontal cortex (PFC), posterior

STS and temporal poles is thought to support theory of mind

processes ( Frith and Frith, 2003 ). Investigations which have

contributed to these conclusions typically involve carrying out

tasks which require inferences to be made, whilst undergoing

fMRI or PET scans. Studies which aim to localize function are

subject to similar sorts of problems as is the task of localizing

mirror neuronactivity. These studieswill involve different brain

processes and regions as they employ a range of experimental

approaches. For example, some use written stories ( Fletcher

et al., 1995; Vogeley et al., 2001 ) and others use comic strips

( Brunet et al., 2000; Gallagher et al., 2000 ) to target ToM

processing. This is an important discrepancy as the involve-

ment and/or requirement of language in ToM is debatable.

‘

BRAIN RESEARCH REVIEWS 54 (2007) 286 – 293

289

Previous work has indicated that temporal language proces-

sing areas are involved in ToM ( Baron-Cohen et al., 1999; Brunet

et al., 2000 ). Other approaches indicate that language is not re-

quired for ToM. For example, children with specific language

impairments ( Perner et al., 1987 ) and even severe aphasic

patients ( Varley and Siegal, 2000; Varley et al., 2001 ) have been

reported to have normal ToM processing. This heavily implies

that language capacity is not an essential requirement for ToM.

Furthermore, language processing areas have also been impli-

cated strongly inmirror neuron studies. Area F5 in themacaque

is thought to be the non-human primate analogue of Broca's

area ( Petrides and Pandya, 1994 ), an area where a number of

studies have reported mirror activity ( Grezes et al., 2003;

Iacoboni et al., 1999 ).

Areas implicated in both theory of mind andmirror ( Table 1 )

include superior temporal sulcus ( Baron-Cohen et al., 1999 )and

parts of the frontal gyrus ( Fletcher et al., 1995; Iacoboni et al.,

2005; Rizzolatti et al., 1996a,b ) and temporoparietal junction

( Williams et al., 2006 ). These are extremely rough comparisons

and only provide a preliminary comparison to demonstrate that

an overlap is a possibility. Given the individual differences in

locations of brain structures and landmarks ( Amunts et al.,

2004 ), firmconclusions in either direction cannot be drawn from

this extremely coarse comparison and further investigation is

required. It is important to note that the majority of the studies

which have investigated ToM have not differentiated between

different types. For instance, inferring the intentions of others

There is a clear distinction between mental simulation

of action, actual simulation of action and understanding

the intention behind an observed action. There are the-

ories which claim that ToM relies on mental simulation of

action (Simulation Theory, see Gallese, 1998 ); however, this

is yet to be confirmed. Nevertheless, the evidence supports

a role for the mirror system in the former two of these

abilities.

Thedirectmatchingtheoryofimitationassertsthat

observation of an action produces an internal motor repre-

sentation in the brain of the observer. The most compelling

evidence to date for the influence of mirror neurons in motor

memory comes from studies by Stefan et al. (2005) .Ina

previous study, this group demonstrated that motor practice

of bidirectional thumb movements influences the direction of

subsequent cortically TMS evoked thumb movements ( Clas-

sen et al., 1998 ). In other words, if you practice moving the

thumb to the left any ensuing evoked movements are more

likely to be made to the left. These finding have been

replicated in other human ( Pascual-Leone et al., 1995 ) and

animal studies ( Kleim et al., 1998 ). More recently, Stephan

et al. (2005) have described how the mere observation of

thumb movements has the same effect on influencing the

direction of subsequent thumb movements. This provides

evidence that action observation influences the neural circuits

responsible for action execution in a positive rather than

inhibitory manner. Given that the mere observation of an

action can trigger an internal motor representation of the

same action, it is possible to see how mirror neurons may

contribute to imitation ( Jeannerod, 1994 ).

Human imaging studies have demonstrated that left

inferior frontal and right superior parietal cortices are

activated in response to both imitation and observation

( Iacoboni et al., 1999 ). These data provide evidence to support

the direct matching hypothesis and indicate that mirror

neurons may be involved in the underlying mechanism.

Furthermore, PET studies have shown that observation of

actions with a view to later imitating the action activates

motor areas involved in generation of actions and planning

( Decety et al., 1997 ).

To conclude, according to simulation theory, we mentally

mimic actions that we observe allowing us to infer the internal

states of others such as their intentions. The evidence for

mirror activity in formulating an internal motor representa-

tion of an observed action in order to imitate thus corresponds

with a simulation theory of ToM. At first glance there appears

to be aminor correlation between some of the areas thought to

display mirror activity and those involved in theory of mind

processing. However, clear characterization of motor ToM and

further studies of mirror activity in these brain regions is

required before any conclusions can be drawn. A more fruitful

motor theory of mind

”

Brain areas involved in theory of mind and those

displaying mirror activity

ToM

–

MNs

fMRI STG amygdala PFC ( Baron-

Cohen et al., 1999 )

Left frontal operculum

(Broca's area) Right anterior

parietal area ( Iacoboni et al.,

1999 )

ACC Left temporopolar cortex

( Williams et al., 2006 )

Posterior inferior frontal

gyrus ( Iacoboni et al., 2005 )

Right parietal lobe ( Williams

et al., 2006 )

Right anterior parietal

cortex ( Iacoboni et al., 1999 )

PET Left medial frontal gyrus

Posterior cingulate cortex

Right posterior STS ( Fletcher

et al., 1995 ) (Verbal task)

Left inferior frontal gyrus

(BA 45) and left STS

( Rizzolatti et al., 1996b )

Left medial frontal lobe Left

temporal lobe ( Goel et al.,

1995 )

Left inferior frontal gyrus

(BA 45) and left STS, left

parietal lobe, right

dorsomedial motor cortex

and mesial area 6 ( Grafton et

al., 1996 )

Right PFC Right ITG Left STS

Cerebellum AC MTG ( Brunet

et al., 2000 )

Functional magnetic resonance imaging and positron emission

tomography have been used to investigate which regions of the

brain are involved in theory of mind processing. A separate range of

authors report areas of the brain which demonstrate mirror

activity. Overlap between areas of the brain which are involved in

processing both would provide evidence for the role of mirror

neurons in theory of mind.

maywell involvemechanisms which differ from those involved

in inferring the beliefs of others.

The most famous test for ToM is one proposed by Dennett

(1978) : the false belief task is designed to test if one is able to

infer the beliefs of others when they are incongruent with

ones own beliefs. Whilst this is an elegant way to investigate

the question it was designed to answer, it is not ideal for

identifying regions of the brain involved in inferring inten-

tions of others, or

‘

Table 1

290

BRAIN RESEARCH REVIEWS 54 (2007) 286 – 293

approach may be to target the mirror system and ToM

processing within the same experiment.

cortex (IFC), inferior parietal lobe (IPL) and STS ( Hadjikhani et al.,

2006 ). These conclusions rely on the previouswork outlining the

location of the human mirror network being confirmed.

An additional group of people with attenuated or absent

ToM processing are healthy children. According to Piaget's

theory of development, until the ages between 6 and 11,

children are unaware that others have their ownmental states

that differ fromour own ( Piaget, 2001 ). There is some debate as

to when ToM appears during normal development. Early

studies have indicated that ToM is absent in children below

the age of 4 years ( Perner et al., 1987; Wimmer and Perner,

1983 ). Detailed study of children's use of language and more

simple tasks where false beliefs are embedded in a play tends

to imply that a level of ToM is present as young as 18 months

( Frith and Frith, 2003 ). Thus it would be of interest to

investigate how and when MN activity develops during this

period and how this may relate to the acquisition of ToM.

A final point: it is not understood how the properties of

MNs differ between the human and the macaque. The specific

properties of macaque mirror neurons do not appear fully to

support their involvement in understanding actions, but

macaques are not able to carry out such a task. Examples of

these points are highlighted by Csibra (2005) . In the macaque

model, not all MNs have motor properties ( Gallese et al., 1996;

Rizzolatti et al., 1996a,b ), some MNs respond to more than one

type of observed action and that this action may not be the

same as the executed action (see Rizzolatti and Craighero,

2004 ). Thus he stresses MN firing would be misleading if they

were solely responsible for interpreting the observed actions

of others ( http://www.interdisciplines.org/mirror/papers/4 ) .

Accordingly, if the human mirror system is involved in ToM,

the nature of the system is likely to have advanced or adapted

in some way from that of our common ancestor. For example,

it is possible that there is an interaction between MNs and

other information in the brain which allows humans to make

these inferences, such as language or memory. The nature of

these changes is not currently understood.

To summarize, data from a variety of fields supports the

suggestion that the mirror system may contribute to under-

standing the behavior of others. In the absence of direct

evidence, this hypothesis needs to be addressed from a

number of stances. Studies which directly target the mirror

system and ToM processing are needed. There are a range of

human groups in which the mirror system can and should be

investigated. Such exploration would help to illustrate the

contribution of the mirror system to mental simulation of

action, imitation and ToM. We have yet to characterize fully

the mirror system in human ontogenesis and phylogenesis.

Such an approach will no doubt provide valuable insights into

how the macaque and human brain functions differ.

3.2. Targeting areas involved in understanding action

intention

If mirror neurons were involved in inferring intentions from

actions, they would be expected to fire in situations when the

intention was not known and thus has to be inferred. Umilta

et al. (2001) recorded from mirror neurons in macaque F5 and

demonstrated that a subset of these mirror neurons fired in

response to the action even when the completion of the action

was hidden from view. Thus the mirror system appears to be

encoding the inference of the intention behind the action. This

carefully designed study comprises a paradigm which targets

both the mirror system and ToM processing in the same time

period. Consequently, this has provided the strongest piece of

evidence for the role of mirror neurons in theory of mind so

far.

Iacoboni et al. (2005) took the next step and focused in on

context within an action in order to elucidate the brain areas

involved in inferring an intention. They found that actions

within a context specifically result in activity in inferior frontal

gyrus and ventral PMC, areas which reportedly demonstrate

mirror activity. The authors interpret this as evidence for the

role of a subset of mirror neurons in extracting information

about action intention from environmental contextual clues.

Together these two studies provide powerful evidence for

the role of the human mirror system in understanding the

actions of others, both in the macaque and human brain. How

this ability differs between the two models remains to be seen.

In order to assess the impact this has on understanding the

behavior of others it is necessary to look to people who have

attenuated theory of mind and/or mirror systems.

3.3. Mirror activity in people with absent or attenuated

ToM

Some patients who suffer from abnormal ToM processing also

have attenuatedmirror responses. Initial studies have focused

on the classical conditionof abnormal ToMprocessing: autism.

The last few years have seen a cluster of studies indicating

abnormal mirror activity is present in autism ( Dapretto et al.,

2006; Ramachandran, 2001; Williams et al., 2006, 2001 ). It has

been suggested that mu-wave activity may be an indirect

measure of mirror neuron activity ( Muthukumaraswamy et al.,

2004 ). Oberman et al. (2005) report that the normal mu

suppression that is seen during action execution is reduced

in people with autistic spectrum disorder. Theoret et al. (2005)

report that modulation of activity in primary motor cortex (MI)

of patients with autistic spectrum disorder (ASD) is reduced

with respect to controls during finger movements. Again

however, we must note that these studies have not used

object-related goal-directed movement and thus may not

reflect the mirror activity. This is not to deny that there may

be additional motor abnormalities in these patients.

Theremay be preliminary structural evidence to support the

claim that MNs are involved in autism. A recent article

suggested that there is cortical thinning in areas reported to

show mirror activity in autism, namely the inferior frontal

4. The role of mirror neurons in different types

of theory of mind

The specific typeof ToMthatwe have describedabove is amotor

ToM referring to the understanding of the intentions of others'

actions. However, ToM is a broad term which covers a range of

cognitive abilities, including understanding of intentions,

understanding of beliefs and understanding of goals. It is likely

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: