Hesitations and intentions in action

We are able to recognize/understand someone else's believes, states of mind,
and intentions. This ability grossly subsumed under the label 'mentalizing'
(forming a Theory of Mind) is thought to form the core of our social skills.
Previous studies have found that autistic individuals often have difficulties
in tasks requiring to understand the believes of someone else (typically, false
believes tasks). This is probably one of the most common deficits found in the
autistic population along with poor eye contact and stereotyped repetitive
behaviours. However, the late achievement of this skill in the healthy
population, around 4 years old, does not match the early time course of
autistic symptoms. The origin of the difficulty understanding other people must
lie somewhere else, in a skill that autistic children fail to develop between 1
and 2 year-old. One hypothesis is that a deficit of the mirror neuron system
could be involved at an early stage of the pathology.

Mirror neurons are neurons from the ventral pre-motor (vPM) and
inferior parietal (IP) cortex that fire both during the execution of an action
and during the observation of a similar action performed by someone else.
Neurons with such mirroring property would enable the observer to enact the
action that was witnessed, and through this process plainly understand its
goal. Evidence from intracranial recording in the monkey actually suggests
that the MNS does not merely encode the kinematics of the action performed by
someone else, but the immediate goal of the action. In fact, mirror neurons do
fire even when the last part of the action (the grasp itself) is not visible.
Finally, mirror neurons also fire to the sound that results from a common
action being performed by someone else.
Some experiments even suggest that mirror neurons could play a role
beyond the mere comprehension of the immediate goal of the action, in
understanding the state of mind and the believes of the agent. The results of
one experiment in humans for instance suggest that the activity of the vPMC is
modulated by the context of the action (witnessing someone grasping a cup for
drinking activated the vPMC more than viewing the same action in the context of
cleaning up the table).
Evidence of a dysfunction of the MNS in autism is meagre. A couple of
studies have found abnormalities in MNS functioning, but others have stressed
the fact that autistic individuals do not show the expected profile of
performance of someone with a MNS deficit. In particular, autistic individuals
are quite able to imitate and to learn by imitation, they sometimes demonstrate
echolalia (repeating what they heard word-by-word with little comprehension),
and may develop exceptional manual abilities (like sculpture, craft, drawing,
or the ability to play music instruments).
On the other hand the hypothesis that the MNS could by it-self serve
the comprehension of higher levels of intentions, believes, and states of minds
has also been criticized. One possibility however is that of a collaboration
between the MNS and the medial prefrontal cortex (MPFC), an area that plays a
central role in thinking about self and others.
Our study is designed to explore the relationships between the MNS and
the MPFC which we suspect are disturbed in autism spectrum disorders.

Our experiment will assess the respective role of the mirror neuron system
(MNS) and theory of mind areas for understanding someone else*s immediate
intentions and state of mind (prior intention). We want to test the hypothesis
that making inferences about the prior intentions of someone on the basis of
the kinematics of the action of the person requires a collaboration between the
vPMC (MNS) and the MPFC (Theory of Mind area). In contrast, making inferences
about the immediate goal of the action of someone is not supposed to require
the intervention of theory of mind areas. The mirror neuron system is likely to
be sufficient to understand the immediate intention of the actor, even when the
goal of the action is out of sight. We ought to show that autistic individuals
demonstrate preserved activation of the mirror neuron system during processing
of immediate intentions, but that the pattern of connectivity between vPMC and
MPFC during processing of prior intentions is disrupted, and the degree of
disruption at this level correlates with the performance of the patient in
theory of mind tasks and perspective taking.

For the present study we have built an original set of stimuli that can be used
to investigate both the process of understanding someone else's intentions in
action, and states of mind. The stimuli are movies of an actor picking a ball
inside a box.
The actor sits behind a table. Only the torso is visible; the head is outside
the field of view of the camera. The camera shows the hand of the actor until
it disappears inside the box. There are always 2 balls inside the box; one big
and one small ball. The two balls always have different colors and different
saturations. On some occasions the actor was told which ball to pick (the big
or the small one). On other occasions the actor was asked to select the less
saturated ball. In this later condition, the actor was likely to hesitate. Some
of these hesitations are indeed very obvious, even on a 2-second movie.
Witnessing an hesitation gives a direct information about the state of mind of
the agent.

Twelve actors were involved in the creation of the stimuli, mostly colleagues
and family members. All the actors were explained the experiment and signed an
informed consent. The actors recruited are between 6 and 55 year-old, half of
them are males. We decided to include actors from different ages and gender in
order to increase (1) the diversity of the kinematics -and therefore the demand
on the simulation mechanism (MNS), and (2) the diversity of the characters -and
therefore any demand on the mechanism of perspective taking (Theory of Mind).

The same movies can be the object of different questions: One that taps the MNS
(e.g. Does s/he pick the big or the small ball?), one that requires mentalizing
(e.g. Was s/he asked to pick the less saturated ball?), and two control
questions, one of disposition attribution (e.g., Is s/he a very smart person?),
and one low level visual condition (e.g. Is the tip of the chin visible on the
video?). Two additional sets of movies were also taken with every actor for use
as control conditions. One series of movies show the actors performing a
non-meaningful gesture above the box (meaningless gestures condition), and one
series of movies show the actor picking a ball on the table with the box on the
side (full-vision condition). This last condition will permit to control for
possible differences with previous studies where the goal of the action is
always visible.

During the fMRI experiment, both autistic and control participants will perform
a number of different tasks while their cerebral activity is recorded in a 3.0T
scanner with standard image acquisition parameters.

1. Execution task
One task is the execution of grasping movements. The results of this task will
be used to delineate the brain regions involved in motor execution of grasping
action. It is a standard procedure to delineate the plausible location of the
MNS. The participants will be requested to pick a ball next to them when they
hear a tone, and to put it back before the next tone (8 sec later). There are
two different tones. Upon hearing a low tone the participant will be requested
to pick the ball on the left, and upon hearing the high tone s/he will pick the
ball on the right. Participants will pick 15 times a ball on each side. Items
are separated by a 12 sec rest period. A total of 120 brain volumes will be
acquired during a single 11-min EPI sequence.

2. Theory of Mind task
Another task will be used to localize brain areas involved in theory of mind
reasoning. In this task, participants will read 24 short narratives about the
formation of a representation (12 about beliefs, 12 about physical
representations like a photo, drawing, or map) that do not correspond to
reality. Stories are on average 32 words long, and are presented for 10s.
Participants then answer a fill-in-the-blank question about the representation
(presented for 4 s). Stories of the two conditions alternate with a 15-s rest
period after each story. Participants will perform two 6-min runs of this
experiment (12 stories each).

3. Action monitoring
In order to identify brain areas involved in action monitoring participants
will be given another localizer. In this experiment participants will perform a
numerical interference task in which they have to indicate the largest of two
numbers that vary in size. Incongruents items (e.g. with the larger number
displayed in smaller font) are mixed with congruent items. Items are presented
in blocks of 5 (with one item every 2.5 s) and the participant answer by
pressing one of two keys corresponding to the side of the response. In a
control condition, letter-number pairs are displayed on the screen and the
participants must indicate the side of the letter. The timing of the stimuli is
the same as for the main condition. Blocks of the 2 conditions alternate and
are separated by a 12-s rest period. Participants will perform two 6-min runs
of this experiment (12 block of each condition).

4. Observation tasks
Finally, in the main task, participants will observe grasping movements and
non-meaningful gestures and will answer different questions about the items.
Items last for 12 sec and are composed of 6 movie clips (about 1.5 s each)
separated by a cross. The same actor is depicted performing a grasping or a
non-meaningful action in the 6 clips. Items are preceded by a one-word
question and the participant must answer after the 6 movies (in order to
model-out the cerebral activity related to the key press and the instruction).
There are 6 different conditions:

C1: Attribution of prior intention / state of mind (Did s/he had to choose the
less saturated ball at least 3 times or not?)
C2: Attribution of immediate intention (Does s/he pick the big ball at least 3
times or not?)
C3: Attribution of dispositional state (Is s/he a very smart person or not? -
or is s/he very self-confident or not?)
C4: Visual control (Is the tip of the chin visible on the video at least 3
times?)
C5: Meaningless (Does s/he touches the box at least three times or not?)
C6: Full-vision control (Does s/he pick the big ball at least 3 times or not?)

The order of condition is randomized. There are 12 items per condition (72
brain volumes), and the number of yes and no responses is fairly distributed
across conditions. Every item will be followed by a 10-sec rest period. Images
of the whole brain will be acquired across 3 runs (12 min each). Before
starting the experiment, the participants will view a few un-cut movies
presenting the complete grasping action performed by the actors, and will be
given the opportunity to manipulate a ball of each size. In addition, the
participants will perform a few trials of the tasks.

The action observation, action execution, action monitoring, and theory
of mind conditions will be tested in separate runs to avoid inflating the
variance of error in the cognitive conditions. A full brain anatomical scan
will be acquired in every participant for rendering of the activation maps, as
well as a short anatomy (with the same number of slices and dimension as the
EPI functional runs) for realignment purposes. Finally, diffusion tensor images
will be acquired for analysis of the integrity of the white-matter cortical
tracks. The total duration of the MRI experiment is 85 minutes. On a separate
session, the participants will perform additional tests of Theory of Mind
functioning, as well as an IQ test (see below).

Behavioral tests
In the behavioral tests, participants will view the same short movies with
different actors picking balls inside the box. They will have to judge the size
of the ball being picked, to detect the hesitations of the actor, to judge the
jerkiness of the movement. Later they will also be asked to evaluate the
self-confidence or self-esteem of the actor on a 5-point scale. The 12 actors
are shown separately. There are 12 movies per actor. The items and the
conditions are mixed pseudo-randomly. Responses and latencies are recorded by
the computer. Participants will be given the IRI empathy scale for correlation
with their behavioral performance, and handedness will be assessed with the
Edinburgh inventory. Results to the tests will be used for the analysis of the
fMRI data in order to identify the effect of different dimensions of the
stimuli on the cerebral activity. The total duration of the session is
approximately an hour.

BURDEN: The burden associated with the cognitive tests is minimal. Short
breaks will be inserted between the tasks at the request of the participant, or
if the experimenter feels it is necessary.
For the MRI, the burden is obviously higher since participants have to lie
still in a confined space while performing the task. However our experience is
that most participants with or without ASD can easily make it, and are very
glad to participate. We kept the total duration of the MRI scan under one and a
half hour in order to minimize the burden. In addition, after 45 minutes, the
participant will be asked if s/he requires a 5-min break.

RISK: The experiments will not entail more than minimal risk to the
participants.

BENEFIT: The study is not intended to benefit the subjects directly. However,
the data collected during this study could improve our understanding of autism
and human cognition at large. In particular, this investigation could help us
clarify why autistic individuals are unable to recognize the state of mind of
someone else.