| |
Introduction
Autism has
played a strong ancillary role in many debates concerning social
cognition, how it develops and its structure. It is thought that, since
persons with autism lack the basic abilities to think about others in a
“social” fashion, understanding autism will give us a window into
understanding much or all of social cognition. Analogous to the role
lesion studies and other neuropsychological disorders play in our
understanding of cognition, autism presents a case in which valuable
insight might be gained about social cognition. Such views are
commonplace in current philosophy of psychology as evidenced in the
following statement “… [studies on autism are] fertile testing-ground
for a whole range of specific hypotheses concerning normal theory of
mind” (Carruthers, 1996).
This paper
will argue that three of the main competing explanations of autism, the
Theory of Mind approach, The Executive Control Approach and the
Simulation Theory, are flawed. This conclusion is based on recent
findings about autism which point to the fact that there are certain
perceptual deficits that persons with autism face. These new findings
show that persons with autism have basic perceptual deficits that
prohibit the development of socio-cognitive abilities that allow them to
understand the behaviors of others. Furthermore, these deficits seem to
be ones that are difficult to account for on certain current theoretical
approaches to our development of socio-cognitive abilities. In addition
to this critical work, this paper will briefly lay out a plausible view
of autism and show that autism still might tell us interesting things
about the development and structure of socio-cognitive abilities.
On this
view, taking greater account of the basic perceptual inabilities current
research is finding in autists confers certain benefits. First,
realizing that autism might not be the result of a certain upper level
cognitive deficiency will allow researchers to blend together insights
from a variety of theories of cognition and cognitive development (many
critiqued in this paper). Second, this view will allow for a broader
range of input from other disciplines such as neuroscience, which will
allow for the addition of possible biological markers for the diagnosis
of the autistic disorder. Much in the same way that current researchers
(Bloom and German, 2000) interested in Theory of Mind (ToM hereafter)
find the false belief task no longer to be the “test” the passing of
which demonstrates a full-blown ToM, we will see that autism is likely
not a disorder that is strictly a metacognitive, ToM or simulation
failure.
The structure of the paper is straightforward. In section 2 we review
some new findings on the perceptual abilities of persons with autism and
review the general reception work of this nature has received in
cognitive science and the general views the research supports. In
sections III-IV, the paper presents an explication and critique of three
dominant views of cognition and cognitive development. The three views
are the Theory of Mind approach, the Executive Control Theory and
Simulation Theory. In section VI, having completed the critique, the
critical work of the prior three sections is brought together to briefly
lay out a new view of autism and some of the benefits such a view has.
New Findings
on Autists
Since autism’s
classification over 50 years ago, autism’s primary diagnostic features
remain focused on the pervasive difficulties autists have in
understanding the social world. Many theoretical approaches emphasize
that these socialization problems are related to high level cognitive
difficulties with thinking about other minds. Some theories support the
notion that persons with autism have insufficient cognitive skills to
manage complex thoughts about the world, others take autisms deficit to
be in an improper simulation of other’s thoughts which supports the
development of our thoughts about other minds. No one theory is complete
and each has its own difficulties.However, recent studies on autism
reveal a number of lower-level perceptual deficits in perception of the
social world. Specifically these studies imply that there are two
perceptual modalities with which people with autism show deficiencies,
the perception of “social” stimuli in both the auditory and visual
modes. Other related research also shows that persons with autism do not
process facial stimuli in the same area of the brain as do normals.
Blake et al. (2003) show that
the perception of biological movement in persons with autism is impaired
relative to controls. In these experiments the children were trained on
light displays showing biological and non-biological movement. Once
children were able to consistently recognize the two classes, they were
then tested on fifty 1-second displays. The control task was to identify
a shape moving among distractor items. Persons with autism were shown to
be impaired on the biological movement task relative to controls while
showing no deficit in the control task. The experimenters conclude that
these results support the view that autism affects a specific set of
neural mechanisms (presumably located in the superior temporal sulcus)
dedicated to the processing of biological movement and intentions. In
another study investigating audition, Ceponiene, et al. (2003)
found that persons with autism were unable to orient and focus on
“speech” sounds.
They found that
persons with autism had difficulties in attending to “speech” sounds as
opposed to no difficulties in perceiving non-speech sounds.
Experimenters ruled out basic auditory problems such as hearing loss and
inability to attend to complex, non-speech sounds. They also ruled out
issues of complexity (as speech sounds tend to be complex) through the
use of counterbalanced presentations that were auditorily complex
without having speech-like qualities. In each trial where speech sounds
were counterbalanced with equivalently complex non-speech sounds persons
with autism responded well to the latter and not to the former. From
these studies they conclude that there is/are specific deficits that
persons with autism have in perceiving “speechness” quality of the
sounds. However, it should be noted that high-functioning persons with
autism were able to form representations of whatever sounds they were
listening to, speech or non-speech, and respond to changes in those
sounds. The problem for persons with autism is that these changes of
response were termed “conscious” changes in attention. What persons with
autism lacked was the ability to subconsciously attend to differing
speech sounds. The authors suggest that this inability is the result of
post-sensory processing difficulties which likely explains the
difficulties many persons with autism have with language development. In addition to the above
findings, there has recently been work done on how persons with autism
process faces: an important part in the recognition of others. The newer
research shows that persons with autism process facial stimuli in areas
outside of the fusiform gyrus: an area where facial stimulus processing
normally occurs (Schulz, et. al., 2000; Pierce, et. al. 2001). The
conclusion researchers draw is that the neural processing of facial
input is not as fully developed in persons with autism compared to
normals.
There is also research focusing on the normal development of
socio-cognitive abilities that seems relevant to understanding how our
socio-cognitive abilities. In a recent special issue of the
Philosophical Transactions of the Royal Society (2003), “Decoding,
imitating and influencing the actions of others: the mechanisms of
social interaction,” edited by Chris Frith and Daniel Wolpert,
researchers were brought together in order to find the neural mechanisms
of social interaction. This issue attempts to synthesize and present new
research on social cognition and its relation to the brain. The issue
encompasses researchers working with neuroimaging data, mathematical
modeling of intentional action perception, understanding the role of the
“mirror” neurons in imitation and various other issues. While a
comprehensive theoretical unification of these findings has yet been
offered, broad agreement can be found on the type of conclusions reached
in the Blake et al. (2003) study. Researchers such as Puce and Perrett
(2003) have hypothesized that recent evidence supports the hypothesis
that there is a specific neural mechanism for the visual perception of
biological movement. Coupling these findings to the research mentioned
above then we might have good reason to believe that humans have special
perceptual abilities regarding the social world. Again, the general
belief is that we are, in a very real sense, wired to perceive
biological stimuli.
Since the general view is that there is some basic sense in which we are
hard-wired to perceive the social world and persons with autism have
difficulty with these basic abilities, we could plausibly hypothesize
that autism is the result of deficits in attending to basic social
stimuli. Such a view then conflicts with recent theorizing about autism
in which persons with autism are hypothesized to have some defective
module for processing ToM input, or a difficulty in simulating others’
cognitions, or in adequately maintaining the proper arrangement of ideas
during recall. As a result, many theoretical treatments of autism, such
as ToM, Simulation Theory or Executive Control Theory might be
misguided. Furthermore, autism might not be a case from which one could
adjudicate between the abovementioned theories of social cognition as to
which is more the more accurate story of socio-cognitive development.
While this is certainly implied by the above evidence, we must turn the
promised critical work of current theories of autism in order to better
understand the role autism is taken to play in understanding
socio-cognitive development.
The ToM
View
In this section we
will review the ToM approach to development and its treatment of autism.
This section relies specifically on the work of Simon Baron-Cohen (1995)
and his theory of autism. However, the phrase “ToM approach” is used as
a general marker for that family of theories that takes our knowledge of
other minds to be innate and basic (See Carruthers, 1996 and Botterill &
Carruthers, 1999 for related ToM views on development and autism). We
first detail Baron-Cohen’s position on ToM
development, give his view of the autistic disorder and then turn to
some problems with Baron-Cohen’s position.In order
to better understand the type of socio-cognitive knowledge approaches
such as the ToM, Executive Control and Simulation Theory attempt to
explain let us begin with an example. Imagine two close friends have
just come back from a night of trick-or-treating one Halloween and have
commenced to surveying the candy we received. Sam, being an aficionado
of hard candy, begins to gather all those types of pieces into a pile.
Sam’s compatriot, on the other hand, is a connoisseur of chocolate and
he recalls them saying so and is reminded of this as he can see her
collecting all the chocolates into a pile. As Sam separate out the
candies from one another Sam mentions to his friend Alice that he would
be willing to trade his chocolates for their candies. This interaction
depends upon the one person representing to themselves the preferences
of another. This would be the sort of knowledge that the theory in
theory of mind allows one to develop. Sam knows that Alice likes
chocolate. Alice knows that Sam has chocolates and might be willing to
trade might be plausible. Note also that the situation requires that Sam
be aware that Alice knows he has chocolates and be willing to trade
them. As shown in the prior example, understanding and recognizing the
preferences, desires and beliefs of others plays an important role in
our interactions. The basic concepts of belief, intention and desire
play the roles of theoretical entities that we postulate in order to
explain behavior. The reason that Alice is separating out her
chocolates, Sam understands, is the result of his belief that she likes
them. The unseen desire, Alice-likes-chocolates, explains the
experienced candy sorting behavior. Baron-Cohen (1995), for example,
believes that our ability to mindread (his neologism for socio-cognitive
knowledge) is the result of four separate modules/mechanisms working
together in order to produce beliefs about what others know. The
mindreading system is broken down into the following four modules, ID-
the Intentionality Detector, EDD- Eye Direction-Detector, SAM- the
Shared Attention Mechanism, and the ToMM-Theory of Mind
Module/mechanism. He believes that each of these four mechanisms line
up, roughly, with properties in the world, which are: volition
(desires), perception, shared attention and
epistemic states (knowledge and belief).
Baron-Cohen first describes the Intentionality detector (ID)
(Baron-Cohen, 1995, p. 32). This is the first mechanism that one needs
in order to mindread. It is a perceptual device whose job it is to
interpret the motion of objects in terms of primitive volitional mental
states like goal and desire. This could mean something like “Alice wants
the chocolates” from the above example. A more general rendering would
be “Object wants/desires x”. Humans use this because it makes sense of
basic animal behaviors like approach and avoidance. In order to
interpret motion in this way, one needs only two conceptual states, want
and goal. The function of the ID is fairly simple. The ID is activated
whenever there is any perceptual input that might be identified as an
agent. We also interpret certain stimuli in the modality of touch as
well as other modalities and sound in an intentional fashion
(Baron-Cohen, 1995, p. 36). If we back up into something we may take it
to be a person, and thus say ‘pardon me’, only after we verify that it
is not a person do we look around to make sure no one was watching us
talk to no one in particular. When we hear the famed bump in the night
we immediately seem to believe that this noise was made by something.
The second device is the Eye Detection Device (EDD) (Baron-Cohen, 1995,
p. 38). The EDD works only through the visual sensory mode. It has three
functions; it detects the presence of eyes or eyelike devices; it
computes which direction the eyes are pointing, and it infers from its
own case that if another organism’s eyes are directed toward a thing,
then it sees that thing. It is important on Baron-Cohen’s view that the
third function be seen as giving the organism with the EDD the ability
to posit mental states about the organism it is viewing. Now a new
mental state, ‘one of knowing or believing that some other creature may
have visual access to’ is added to the basic/primitive mental states of
the child. The second and third functions of the EDD are important for
Baron-Cohen. Baron-Cohen believes that it is highly adaptive to be able
to make a judgment about another being’s knowledge, such as when the
tiger has prey in its sights (see Baron-Cohen, 1995 pp. 32-36). If one
calculates that the tiger has its eyes trained on a friend, and one uses
their knowledge that eyes are used to see (extrapolation from self and
third function of the EDD), then one should realize that the tiger sees
one’s friend and probably will want to attack. This is called a dyadic
representation: Agent sees X. The ID and EDD can form dyadic
representations which are relations between two objects or people. It
resembles the story told about the tiger. With the ID one can interpret
the tiger as an agent. If the agent sees ones friend, and eating is a
desire of the tiger, then one might realize that my friend is in danger.
The third mechanism we will
deal with is the shared attention mechanism, or SAM (Baron-Cohen, 1995
pp. 44-50). The SAM is a mechanism that has as its sole function
building triadic representations. The triadic representation expresses a
relation between object, Self, and agent. The representation is put
generally thus: [I-see- (tiger-sees my friend)]. The SAM compares input
from the ID and the EDD and forms these triadic representations.
Continuing with the tiger example, with a slight modification, will
help. If one sees the tiger prowling (ID) and then looks at your friend
some yards away and one also notices that the tiger sees your friend (EDD),
the SAM can now extrapolate that both the tiger and you see your friend,
because the tiger and myself are both now in a place where we can see
them. Furthermore, say you know that tigers like to hunt humans; you
might then warn my friend of his impending lunch date. In this scenario
the SAM makes available the ID’s inference that the tiger has a goal,
which one interprets through experience, to the EDD and then reads the
eye direction in terms of the agent’s inferred goals. With this
information one might surmise, according to the example, that the tiger
would, more than likely, eat my buddy. After reaching this conclusion I
may yell to try and warn your friend of her danger. With all of this in
place we can see that this use of primitive representations could be
very adaptive and helpful in navigating through a world that has agents
some of which act with goal directed activity. The final mechanism in
Baron-Cohen’s architecture is the Theory of Mind Module/Mechanism (ToMM)
(Baron-Cohen, 1995 pp. 50-55). The ToMM has a number of distinct
functions. Baron-Cohen believes that the ToMM is a cognitive system that
allows the human the ability to posit a wide range of mental states from
observed behavior— to employ a theory of mind in parsing the behavior of
others.
We learn that upon seeing a desired item, ceteris paribus, people
will likely try to get that item. We also learn that people can often
misrepresent the world and that these false-beliefs might lead to
behaviors that are explainable only in terms of this type of a false
belief. The ToMM is the one mechanism/module that we can utilize in
order to understand and codify what we learn about mental/epistemic
states. The ToMM gives us the ability to represent epistemic states.
These epistemic states include believing, pretending, and dreaming. The
final responsibility of the ToMM is be able to put the various epistemic
states together in order to allow us to understand how these pieces work
together in mental life. The ToMM has a grand job according to
Baron-Cohen: “It has the dual function of representing the set of
epistemic mental states and turning all this mentalistic knowledge into
a useful theory” (Baron-Cohen, p. 51). The ToMM has multiple functions.
It first processes representations of propositional attitudes of the
form: [Agent-Attitude-“Proposition”]. An example is “Selma believes that
it is wintery.” Alan Leslie calls these M-representations. This
is a different ability than having a mental representation of, “It is
wintery today”. It differs because my belief about Selma is a
representation of what I take her to be believing about the world.
Having these sorts of representations is crucial to the ability to
represent epistemic mental states. The ToMM also allows us to infer that
a person will attempt to obtain what they desire if they believe that
they are likely to succeed. For many ToM researchers, autism, and the
problems persons with autism show on a variety of ToM tasks, is often
taken as evidence for the innate basis of our cognitions about
other minds. For example, persons with autism do poorly on the
false-belief task.
Persons with autism typically use less mental state attribution in their
speech relative to normals and IQ matched developmentally delayed
children. Persons with autism also fail to recognize surprise based
emotions in others (Harris, 1989). However, persons with autism do show
preserved cognitive function in areas as diverse as mathematics, music
and mnemonic capacities. These preserved cognitive abilities in persons
with autism, it is argued, support a dissociation which furthers the
case that ToM knowledge is separate, and thus likely etiologically
different, from other cognitions. Often this dissociation of cognitive
functions stands in favor of viewing ToM cognition as different from the
rest of cognitive function. That is, ToM represents a domain of
cognitive function.
The use of “domain” here should be seen as heuristic rather than
pointing to any sort of natural kind.
Furthermore, a case for double dissociation can be based on the fact
that mentally retarded children who have lower cognitive function than
normals as well as have delayed motor development maintain relatively
intact social abilities. Williams Syndrome children have impaired
spatial knowledge development while they have heightened social
interaction abilities (Karmiloff-Smith, 1992). Cases such as these
provide the necessary contrast classes to secure the double dissociation
needed to conclude that autism is affecting a specific area/domain of
cognition. A domain is a specific class of stimuli to which a
module/mechanism might respond. Traditionally, in sight you have color
perception for the analysis of shape and some “higher-order” domain of
the facial features of conspecifics. In the auditory domain we might
have a specific domain of human utterances. This rather loose set of
examples, (borrowed from Fodor, 1983, p. 47) is only a nod towards a
domain. In any modular system a domain is defined as “that set of
stimuli that induces the input analyzer (a.k.a. module) to fire
completing the type to token computations that are needed” (Fodor,
1983). So, for the autist, the domain of the ToM stimuli is somehow
ignored or misrepresented due to a defect in the system we have for
mindreading.
For Baron-Cohen, the autistic
deficit arises within the SAM. There is a “massive impairment” in the
functioning of the SAM (Baron-Cohen, 1995 p. 66). The impairment of the
SAM feeds forward into the ToMM ultimately causing the mindblindness
that we see in persons with autism. The autistic deficits are thus the
result of related processing difficulties in both the SAM and ToMM.
While the ToM view of autism is interesting and has sparked a good deal
of research, it is not without its problems. Some of the issues facing
ToM are internal in that the theoretical framework of modularity it uses
as a background for the ToMM seems unsuitable (see Fodor, 2001 pp. 55-78
for an extended critique). Of the many problems facing the ToM position,
one of the more troubling is its reliance on the false-belief task.
Recent critiques question the importance of the false-belief task in
telling us anything important about ToM development
(Bloom and German, 2000) or autism in general. The false-belief task
has long been one of the main tests of autistic (in)ability. The general
thrust of the critiques is that the false-belief task taps into
developing cognitive abilities that do not use a special type of
knowledge base in order to work. That is, the capacity to process all
the relevant information in the false-belief tasks require a host of
ancillary abilities that develop prior to and independently from social
cognition. It is not a specific capacity to represent knowledge of other
minds that is used in the false-belief task, rather it is the blending
together of more basic abilities such as holding multiple
representations in mind and the additional ability the child has to
maintain more information “on-line” that allows them to recognize false
beliefs. Failure indicates only processing problems related to working
memory storage limitations not ToM difficulties specifically.
While the above critique is important there is another more pressing
trouble for the ToM account. Specifically, there is simply no reason to
expect lower level perceptual problems persons with autism have recently
been shown to have. Many ToM approaches require that lower level
perceptual abilities are intact. The account of autism offered by many
ToM positions takes autism’s root problem to be a higher level cognitive
deficit in forming representations about the epistemic states of others.
According to Baron-Cohen (1995) eye-gaze monitoring behavior occurs
because we know that seeing leads to knowing. The higher level
representation is the cause of the behavior. To allow a basic perceptual
problem to be the root cause makes the explanation proffered by
Baron-Cohen superfluous. In addition to the causal issue of the above
paragraph that the recent studies show, there are some deeper troubles
these perceptual difficulties present for the modular approach
Baron-Cohen’s position relies upon. For instance, any theory of modular
function must have the proper perceptual input in order for the modules
to carry out the functions it is responsible for. Modules compute
type-token relations based on a certain limited set of inputs. Any type
of perceptual deficit would threaten the proper working necessary for
the modular architecture Baron-Cohen espouses. Lack the proper inputs
and the modules cannot compute the type-token relations that they need
in order to give forth the proper outputs. On Baron-Cohen’s account the
basic perceptual functions of the autist are assumed to be, and
furthermore must be, intact. This assumption is simply contradicted by
the findings of the earlier studies. It seems that the basic perceptual
deficits persons with autism have are more far reaching than simply an
inability to monitor eye-gaze as Baron-Cohen espouses. It might be the
fact that the inability to monitor eye-gaze is the result of some of
these earlier perceptual deficits, but that would place the locus of the
autistic disorder in an area other than what is claimed by ToM
adherents. Ultimately, the recent work detailing the perceptual
difficulties persons with autism face seriously undermines the ToM
position.
So, in the face of the evidence the ToM position is left in an
uncomfortable position. Either it must recant on its position as to the
deficits that cause autism, or it must come up with a plausible
explanation of how the SAM, ID, EDD or ToMM (or any other putative
modules) function on improper inputs. If the ToM adherent opts for the
first choice, they would lose their proffered locus of the autistic
deficit. If they take up the latter, they must rework the modularity
theory in a way that is much different than modularity theories are
typically taken to work and this would threaten the very theoretical
building blocks of modularity Baron-Cohen used to base his view. Neither
choice would leave the ToM position unaltered so we must look to a
competing explanation of socio-cognitive development. Such an
alternative can be found in the Executive Control theories of
development.
Executive Control Theory
An alternative to the ToM
view of knowledge and development is known as the Executive Control
theory. This view states that our ability to understand the mental
states of others is the result of the development and use of more
general cognitive and metacognitive processes such as metarepresentation
and monitoring of problem solving.
Metarepresentation is the ability that our minds have to represent a
representation or have beliefs about beliefs. So, on Executive Control,
to represent to myself a belief state of someone else, i.e. “I believe
my friend sees my chocolate is in the bowl”, one does so with the
understanding that one am representing the belief state of another
person. According to the Executive Control view, these highly complex
cognitions require certain cognitive resources which develop over time
and practice. Furthermore, the ability to represent the mental states of
others is not native. The metarepresentation of another’s epistemic
state is the result of applying general cognitive strategies and
abilities within a specific domain.
On the Executive Control approach the mind is a
domain general information processor able to utilize a wide variety of
cognitive resources across a number of domains in solving problems.
Executive Control models of cognition and cognitive development state
that most, if not all, of our upper level cognitive abilities are
subtended by the same basic sets of cognitive resources. Our ability to
pretend, to problem solve and anticipate the actions of others based on
inferred thoughts we take a person to have all stem from basic general
cognitive abilities. We use the same sets of cognitive resources to
solve problems in math, the social arena and learning our own phone
number. Understanding others’ behaviors in a social setting is no more
than
a particular problem that humans must face.
In order to understand this particular arena, we simply use these other
cognitive skills within the social domain. Executive Control models rely
on a traditional psychological division of labor in the mind which
separates memory into long-term memory (LTM) and short-term or working
memory (STM). On the Executive Control theory we also have certain
cognitive abilities such as the development and use of certain problem
solving strategies and the ability to metarepresent. In addition to the
strategies one uses to solve problems, one must also be able to generate
a plan or method of solving problems that one can implement. As such,
one of the mind’s general abilities is the ability to organize and
reorganize activities as a person solves a problem. “Executive function
is defined as the ability to maintain appropriate behaviors such as
planning, impulse control, inhibition of prepotent but relevant
responses, set maintenance, organized search, and flexibility of thought
and action” (Ozonoff, et al., 1991, p. 1083). Since one knows that they
want to be home by 3:00 PM this afternoon, they realize that they must
finish up the writing have scheduled for today. They must also meet with
students. If they realize that student meetings tap their energy leaving
me unsuitable for writing, they must then plan to write before my
meetings if they are to accomplish my goals.
In certain problem solving situations, we are able to monitor our
strategies for result and economy and make changes with these goals in
mind. In the above case, one might simply schedule meetings on days that
one does not intend to write so that they might more effectively write
on the other days. We can also monitor our performance in reaching
certain goals. If it turns out that the division-of-academic-labor plan
is not working, one may then alter that plan. One might even inhibit the
tendency they have to allow other factors of my job to take time away
from writing. If one stumbles onto a procedure that works well in
getting me “primed” to write, one might adopt its use in order to write
even more. There are many tests used to evaluate our executive control
abilities but the problem confronting experimentalists is that it is
often hard to develop a task that reliably taps one set of skills or
abilities. However, there are some direct tests, one of the more famous
of which is the Tower of Hanoi Puzzle, which researchers rely on to test
executive abilities.
In the Tower of Hanoi tests, participants follow certain rules in order
to accomplish the task of moving the stack of discs from one area to the
next. Imagine that you are presented with three poles the rightmost of
which has three discs of differing sizes. The goal is then to move the
configuration of discs you are presented with, largest disc on the
bottom followed by the next smallest on top and then the smallest on top
of that, to the leftmost pole. You are told that while you accomplish
this task you can only move one disc at a time, you cannot place a
larger disc onto a smaller one and that you need to accomplish the move
in the fewest possible number of moves possible. As you might imagine,
initial solutions usually involve mistakes and a great
many more moves than is necessary.
Persons with poor
executive control (children, patients with certain frontal lobe
problems, persons with autism, etc.) typically perform poorly, relative
to normals, on the Tower of Hanoi task. The reason for these failures is
clear, according to the Executive Control theorist. Failure to perform
well on the tower task involves the inability to properly plan a
solution. It also requires that you are able to remember all the
necessary rules that constrain your choices. This task also measures the
inhibition of prepotent responses, the first of which is to just start
moving the discs over to the leftmost pole. Unfortunately, this is not
necessarily the wisest first move. If it is the case that persons with
autism typically do poorer on this task than matched normals, this shows
that they have poor executive control abilities. As we will see shortly,
persons with autism are not necessarily any worse on this task than
other persons of equivalent intelligence.
Other tests of Executive Control function include a variety of card
sorting tasks that require the participant to sort the cards based on
color, shape, category, etc. Participants are not told the rule for
sorting that will be used during the test. They must figure it out as a
result of the response from the experimenter affirming or denying the
given response. For example, a set of cards will have animals and
artifacts that are colored either red or blue. If the rule the
experimenter is using is based on color, the participant, provided there
are no conditions preventing the learning of the rule, will figure that
the proper rule is “like colored cards with like colored cards”.
However, at a certain point during the test, after the participant has
shown they are using the proper rule, the rule changes to another
dimension which requires that we sort the cards according to object type
(artifact or natural object) on the card. In order to succeed in the
task the participant must become aware of this rule change and alter
their responses accordingly. This test focuses on strategy,
perseverance, and the inhibition of prepotent responses and flexibility
of action. As with the Tower of Hanoi puzzle, persons with poor overall
executive control do poorly on such tasks. While the abilities tested in
the Tower of Hanoi and card sorting tasks are certainly necessary for
the development of our understanding of other minds, they do not
represent the full complement of skills required for awareness of the
thoughts of others. There are still other abilities and skills
necessary. On the Executive Control theory, social knowledge comes from
our ability to pretend which allows us to metarepresent. Pretence, for
many Executive Control theorists, is critically important to the
development of metarepresentation (Jarrold et al., 1993). The skills
involved with pretence are exactly the same skills required when we
begin to think about other minds. When we engage in pretence we are able
to divorce the representation of the object from the object itself: the
representation becomes decoupled. This allows children the crucial move
that separates representation from the object. Once this ability is
practiced, the child then realizes that the representation of the object
is
different from the object itself.
Upon the realization
that the mind represents and can have representations about the world
that are not tied directly to the world (pretending the hall runner is a
parking lot for cars) they are then able to metarepresent a variety of
epistemic states. In order to represent to itself the belief state of
another the child must be able to understand that they themselves hold
representations of the world. They further understand that others have
the same types of relations to the world with their thoughts. Children
can then create a metarepresentation of the person who has some sort of
perceptual contact with the world and then based on that
metarepresentation are able to predict what that person would do in a
given situation. For instance, if Sam knows that Alice saw him hide his
candy in the box under his bed, then he could suspect that she might go
get some if she wants some chocolate. Such metarepresentational
abilities also allow us to recognize the so-called “false-belief” states
of others. Sam must be able to recognize that Alice saw him put the
chocolate in the box under my bed, know that he changed the hiding spot
unbeknownst to her and then realize that she wouldn’t know that the
hiding spot had changed since he never saw me move the chocolate. She
would have a false-belief based on his particular epistemic relation to
the word that he realizes to be inaccurate. Understanding that someone
has a false belief also requires that the user must have cognitive
control over the contents of their minds so that they do not confuse
their own thoughts with what they take others to believe. Only after
these ancillary abilities are developed and work properly can the child
succeed on recognizing the false-beliefs of others. Note that these
complex chains of thought require a large working memory span that
tracks not only my wants (to keep the chocolate for myself), but also
the desires and beliefs of another (Alice wants the chocolate and
believes it’s where Sam first hid it).
A result of this particular view about cognition, development and our
metarepresentational abilities is a markedly different approach and
explanation of the disorder autism than we encountered with the ToM
approach. Instead of taking the root problem of autism to be due to a
failure of some mechanism/module dedicated to the processing of certain
social stimuli, the metacognitive approach finds that autism is the
result of an inadequate working memory which allows us to metarepresent
(Keenan, 2000). The autistic disorder is the result of a failing of the
Executive Control mechanism responsible for inhibiting certain
responses, problems in working memory and recall as well as inflexible
and perseverative problem solving strategies (Ozonoff, et al., 1991).
The failure of persons with autism on typical false-belief tasks is the
result of being unable to differentiate their own views from another’s
during recall (Hughes, 2002).
They might also adopt
the improper strategy of relying on their own personal beliefs, either
by confusing which set of beliefs belongs with whom or simply forgetting
which belief is theirs, in answering questions about others’ beliefs.
The problem facing persons with autism and causing their suite of
behavioral problems is thus a general inability to accurately store and
recall information rather than a specific focal deficit in understanding
mental states.
This domain general
view of cognitive development is not without support. The Executive
Control approach takes the superior performance persons with autism show
on the hidden figures task
as evidence. Persons
with autism do remarkably well at finding the hidden figure whereas
normals tend to become distracted by the whole figure and find the task
of finding the hidden figure difficult. Persons with autism ignore the
whole and merely see the parts. That is, they are unable to unify parts
into global wholes. Their working memory is unable to inhibit attention
to the details in favor of a global awareness. In both the hidden-figure
and the false-belief task the requisite metacognitive abilities within
working memory are missing preventing the higher level cognition to
operate. Furthermore, persons with autism show roughly the same types of
problems in card sorting tasks that many patients diagnosed with
executive control problems (Ozonoff, et al., 1991). Many of the problems
persons with autism evidence in card sorting tasks highlight the
problems with problem-solving they have. Problems such as the
perseveration of inappropriate rules for solution of problems and
inflexible solution strategies are all what neuropsychologists typically
take to be indicative of executive control dysfunction. It is
hypothesized that these disabilities play an important role in the
development of the disorder of autism.
However, recently there have been findings which call into question the
extent to which persons with autism actually have the executive control
problems they are usually credited with having. Recent empirical
findings (Ozonoff and Strayer, 2001) have demonstrated that persons with
autism are not impaired in certain metacognitive functions like recall
and other working memory tasks. Jarrold et al. (1994) have found that
persons with autism do not evidence the types of reluctance to pretend
with items in specific fashions (pretending an item is an item
counter-functional to its typical use) as the Executive Control model
would predict.
Furthermore, the
findings of the Blake et al. (2003) study do not support the view that
persons with autism have executive control difficulties. Recall that
persons with autism were able to attend to complex sounds but not
complex speech sounds. If there were Executive Control problems, then
the responses to both the complex speech and non-speech sounds should be
equally as poor. Furthermore, persons with autism were able to pay
attention to global characteristics of non-biological displays. The
person with autism was impaired with respect to a biological point-light
display.
Such a discrepancy would not be predicted on the Executive Control
model. In Ozonoff and Strayer (2001) they looked specifically at
the efficacy of the Tower of Hanoi task as a predictor of working memory
dysfunction in persons with autism. They found that this task did not
show persons with autism to be specifically deficient in working memory
as predicted by the Executive Control model. Thus, the explanation given
by Hughes above (2002) is not correct. Since persons with autism seem to
have the proper working memory to guide them in successful completion of
the Tower of Hanoi task, it seems unlikely that working memory span is
too small to entertain the thoughts and metarepresentations of other’s
thoughts at the same time. While the authors question the plausibility
that there is some sort of lack of working memory problem at root here,
they do not dismiss the Executive Control hypothesis all together. They
claim that like so many other tests of psychological function the Tower
task might not solely be tapping working memory. It might be that what
is actually being tested in the Tower scenarios is the ability to relate
goals.
In closing this section there is more evidence that seems to tell
against the Executive Control model. It is known that Temple Grandin, a
high functioning person with autism, has difficulty in understanding
social situations (Sacks, 1995). She claims that she simply doesn’t get
them (social interactions) with the ease that normals seem to. However
she is not without her own resources in dealing with social situations.
She claims that she often develops rules extrapolated from memory and
information about social events supplied to her by others. Instead of
intuitively realizing that an inappropriate remark had been made as a
result of picking up on the cues rather immediately, she claims that she
must analyze the events and check to see if they conform with a rule she
had developed prior to the situation. If the experience fits, then she
knows. Furthermore, the application of this rule according to Professor
Grandin is always consciously applied to situations. This is quite
different from the ways in which normal persons are able to understand
cues and social situations quickly and without conscious reflection. The
use of these sorts of rules seems quite a complicated task that employs
a wide variety of executive control abilities such as working memory,
monitoring and goal oriented problem solving. Both the above mentioned
results of Ozonoff and Strayer (2001) and the case of Professor Grandin
threaten the integrity of the Executive Control explanation of the
autistic disorder.
If the above cases generalize to the autistic population at large, then
they do not seem to have the inhibition of response problem that is
often used to explain autism. Since children with autism were able to
find and follow the form among distractors, they are able to process and
monitor global patterns. This means that working memory can inhibit
attention to smaller patterns. These studies bring into question the
view that autism is due to an executive control dysfunction. In the next
section we will see that Professor Grandin’s case poses difficulties for
the Simulation Theory as well. It is to this approach that we now turn.
The Simulation Theory
In this section we will take
a closer look at the Simulation Theory’s approach to our understanding
others’ thoughts, as well as its explanation of the autistic disorder.
This section begins with a brief elucidation of Simulation Theory’s main
tenets. In this section we specifically evaluate Currie and Ravenscroft
(2002) and the approach to simulation found therein. I then turn to
Simulation Theory’s characterization of the autistic disorder and
finally to a critique of this approach in light of problems the theory
faces.
Simulation Theory is an approach to our knowledge of other minds that
blends interesting insights from both of the previous theories evaluated
above. The benefit, to this paper’s project at least, would be that if
Simulation Theory has many of the same theoretical commitments as does
the ToM and Executive Control approaches, then my previous critiques
need only be shown to affect Simulation Theory in the same fashion. If
the critiques are effective against the other theories, then they should
be effective against this one. Unfortunately, on this particular version
of Simulation Theory, its nuances prevent such a move. In fact, this
view proffers the same sort of initial perceptual problems to be
important in understanding autism examined in section I. That is, Currie
and Ravenscroft believe that initial perceptual deficits do not allow
the proper simulation practices to build and develop (Currie and
Ravenscroft, 2002, p. 158). However, according to the authors, the
deficit persons with autism have is not exactly a perceptual one.
Currie and Ravenscroft (2002) claim that simulation allows us to carry
out a certain type of mental process in predicting others’ behavior. In
order to understand what another might be thinking or do, we are able to
project ourselves in imagination into the situation of others. The
nature of this projection is important. In short, one can imagine that
they are walking in the shoes of another, but this is not simply a
perceptual imagining, it is a propositional type of imagining which
contains certain propositions about the person one is imagining. Second,
one can then allow those pretend beliefs to influence what one might
take someone to do. That is, those pretend beliefs play a role in
creating the proper causal connections in thought which allow one to
think they will do x at a later time.
On the Simulation Theory, each person is able to imaginatively project
themselves into the place of another person and “…generate within
ourselves states of imagining that have as their counterparts the
beliefs and desires of someone whose behavior we want to predict”
(Currie and Ravenscroft, 2002, p. 51). Furthermore, all the simulation
takes place in imagination. Our ability to predict others’ behavior
requires that imagination is active in order to run the simulation. Our
imagination provides the mental area in which we can simulate the role
beliefs would play in certain inferential practices of an entertained
person. So, if one imagines that another is hungry, then one might
believe that they will go get lunch. One does this because when one
believes themselves to be hungry they go get lunch. One plugs in
supposed beliefs and desires and then run a simulation as to what these
states would cause them to do in that situation.
In order for one to properly predict based on simulation another’s
thoughts or behavior, certain assumptions must be made. When one simply
thinks “What would I do in this situation” in order to allow the proper
inferential chain to go through, one must assume that myself and the
target they are evaluating are roughly equivalent in a number of
important respects. One assumea that they will avoid pain and that they
would like to further their goals, etc. If one lacks any of these
assumptions, or for some other reasons believe that the person they are
trying to simulate is different in important respects, one must augment
my simulation with this information so as to have accurate predictions
of the other’s behavior. One must disregard or replace certain basic
assumptions that they might entertain in a normal case. Thus, the type
of simulation one must perform becomes more complex. So, in a typical
case, one would predict that their friend, whom they know is hungry will
likely attempt to go get lunch if the opportunity presents itself. One
can make this judgment based on the fact that they would do the same
thing in the situation. One plugs in the relevant information and runs a
simulation. However, if one knows that their friend is on a diet they
have to take that into account when simulating their behavior. One
cannot simply run the simulation using their own particular beliefs, as
they are not on a diet. Details of this sort are crucial in
understanding and predicting behavior.
Simulation Theory, according to its proponents, is better than both the
ToM and Executive Control approaches. It is much more economical than
the ToM (or ‘theory’-theory) approach in that it relieves the knower of
the use of a complex psychological theory. Instead of plugging in
supposed beliefs and desires of some person we wish to understand into a
theory and then churning out a prediction based on our theoretical
musings, we simply imagine what we would do in the same situation. An
example the authors give to support this view comes from the movie
Apollo 13.
In the
film Apollo 13 the team at Houston need to find a way to reduce
the power demands of the stricken craft. They might have tried to do
this by calculations based on premises drawn from the relevant physical
and engineering theories, together with statements of relevant initial
condition. They did it another way: by having an astronaut play with the
power system of a comparable craft on the ground, until he found a way
to run the craft on power known to be available to Apollo 13.
(Currie and Ravenscroft, 2002, p. 54)
While one, as an engineer,
might have certainly run on theoretical convictions to advise the
astronauts, it seems my objective of safely bringing the crew down is
best served through the use of a real ship that might have roughly the
same conditions. For we know that often there is a wide gap between
theory and practice. When presented in this fashion the economy of the
simulation theory is certainly understandable.
Bearing in mind this
sketch of Simulation Theory, let us turn to the view Currie and
Ravenscroft have about the cause of autism. Currie and Ravenscroft
(2002) explicitly note the types of theories there are explaining
autism. They specifically recount some of the main tenets of the same
theories evaluated in the earlier portions of this paper. They note that
what most distinguishes the ToM approach from the Executive Control
model is that that the former is domain specific while the latter domain
general (Currie and Ravenscroft, 2002, p 139). They claim that their
theory is best considered a domain general view. Specifically, they
think that“…autism
be regarded as a disorder of imagination, and we think of imagination as
device which assists us in understanding and solving problems across a
range of different domains. One of those domains is likely to be the
mental: the recreative imagination is heavily employed in helping us to
understand and respond to the thoughts and feelings of other people
(Currie and Ravenscroft, 2002, p. 139).
They further believe that
recreative imagination is an executive device, since, according to
Currie and Ravenscroft (2002), one of the Executive’s functions is to
provide a supplement to reflexive or preprogrammed responses. Recreative
imagination allows us to plan a course of action and work through it
without having to actually undertake the activities planned. If the
aforementioned reasons are indeed the primary cause of autism then there
ought to be other problems, problems in pretence and planning. On
Simulation Theory, autism is the result of an inability to properly use
imagination in the problem solving issues of the mental: specifically
the process of placing ourselves, imaginatively, into the place of
another. However, the problem facing persons with autism is not a whole
scale inability to place themselves imaginatively in the situation of
another. Rather, it is a difficulty in developing the skills necessary
to practice the imaginative replacement.
Placing yourself in someone’s position, as detailed above, requires that
you allow certain belief or desire states that you do not have to become
active. We must set aside our own “mental economy” and allow the
entertained propositional states to guide our beliefs of what that
person might do. As with the earlier example of eating when hungry,
since one is not on a diet, one must set aside their own responses and
think “as if” they were. Thus, one would choose to not eat in the face
of the hunger. Part of the difficulty persons with autism face is they
are simply unable to make the proper adjustments to their own mental
economy to allow the imagined belief states to play the proper role in
simulating another’s beliefs. Persons with autism simply find it too
difficult to simulate another person’s belief or desire states. The
reason for this difficulty is interesting. Currie and Ravenscroft claim
that the reason that persons with autism cannot simulate others is that
they were never able to develop those abilities that allow for complex
simulations to occur.
The reason persons with autism lack the development and use of ToM
abilities is that they lack the “quasi-perceptual capacity for emotion
recognition” (Currie and Ravenscroft, 2002 p. 159). They take the
ability to recognize emotions to be something that is native or that it
surfaces early in development. Since persons with autism do not pick up
on the basic emotional cues, they lack one of the primary inputs that
allow simulation to occur. According to the authors, a young child
perceives another’s emotional state, mimics those facial/bodily
expressions and, based on how that mimicked facial expression feels to
them as they perform it, thereby know what it feels like to be in that
state. Since a person with autism does not even cue into these basic
emotional states, they are never in a position to make the proper
“like-me” reasoning and they never begin the basic mimicry that sets the
whole simulative process into motion. The effects of this simple
inability to recognize and simulate other’s emotional states are
far-reaching. Currie and Ravenscroft (2002) indicate that
simulation is based on and fueled by
very basic perceptual abilities. The infant must be able to cue into
social stimuli.
Once the infant can see these cues they can begin to mimicking certain
features of the emotional expression. Once they begin to mimic the
expression the child begins to generate the affect states involved in
the mimicked display. According to Currie and Ravenscroft, once these
feats are accomplished the infant can assume that if the perceived
creature is in a state, and the infant knows what that state feels like,
whatever they feel is felt by other. The idea is that a very basic “like
me” judgment is made and from that judgment an understanding of others
begins. As the children begin to track eye-gaze and use
proto-declarative pointing, they can then begin to develop more
sophisticated ways of understanding that aids them in understanding and
predicting the behavior of other. Persons with autism are simply unable
to develop the “like-me” judgment when they are tracking eye-gaze or
attending to emotions. Later on, when more complex activity, such as
predicting others’ behaviors based on perceived epistemic states,
emotional cues and desires, a larger operating space is required. The
imagination then provides a setting in which more complex activities,
such as full blown simulation can occur. Since persons with autism lack
these basic abilities they are never in a position to develop the more
complex ability to imaginatively project into another’s situation. So,
far from autism being a problem with simulation, autism is the result of
early difficulties with perception that prevent the proper simulative
abilities to develop.
Thus, autism, for Currie and Ravenscroft (2002), is an imaginative
disorder. There are Executive Control problems like those mentioned in
Executive Control models, but these problems come after and as a result
of the inability to pick up on the basic perceptual content that cues us
in to the mental states of others. While there is much in this view that
is important—the authors take the role of early perceptual abilities to
be important to our understanding of other minds and the relation
metarepresentation has to thoughts about others’ epistemic states—there
are some problems with the view. What are some problems that Simulation
Theory faces? Some of these problems are well represented in the
literature and any interested reader is encouraged to pursue them
further (See Carruthers, 2003 for a review). We also revisit the case of
Professor Grandin and how her understanding of the social world might
not actually require a fully or even partially defined mentalistic
concept in order to be useful.
In a review of Recreative Minds, Peter Carruthers (2003) claims
that there are some deep problems with Simulation Theory. According to
Carruthers, we are not able to adopt contrary to fact desires as the
theory requires. For example, it is easy for us to recognize that a
character might be misled into believing certain events occurred when in
fact they haven’t. The imaginative leap required is not so difficult.
However, Carruthers (2003) points out that it is more difficult for us
to entertain false-desires, or desires that are not our own. While one
can readily imagine someone taking an irrational action based on false
information, it is genuinely harder to simulate a desire that is quite
foreign to me. If there were aliens that desired eating their young,
such a desire is hard for me to simulate. One might understand the
practice as it’s embedded in a certain social structure, but one could
not actually have that desire. The problem for the Simulation Theory is
that it requires that we have proto-desires that we plug into our
simulation. If we are unable to plug them in, then we can’t have a
proper simulation. This, according to Carruthers, seems wrongheaded.
While one may not be able to simulate the desires of others, one can
simulate their beliefs and this might be sufficient to understand and
predict what they might do. A further difficulty is that is emotion
based simulation is an important first step in developing a simulation,
then the inability to simulate desires seems contradictory to Currie and
Ravenscroft’s view.
However, beyond this interesting criticism, it seems that Currie and
Ravenscroft are incorrect that the pockets of abilities that autists
have in social affairs require that they understand mentalistic terms.
They seem to think that whatever mentalistic rule any person with autism
develops requires that they have some basic knowledge of the mental
concept upon which the rule is based. Contrary to this, it seems that
persons with autism could have a pragmatically useful rule of thumb that
explains behaviors, without requiring mentalistic concepts to do any
work. Recall the case of Professor Grandin’s rule following. She notes
that much of the social world, the cues, the movements, the use of
voice, prosody and intonation, escape her. She does not pick up on those
cues. Professor Grandin describes that she has generated rules that cue
her into the behaviors she needs to be aware of to more successfully
navigate the social world. She describes her rule building process as
one based off and extrapolated from experience. While she now knows that
certain responses indicate “offense” or “humor” she doesn’t immediately
recognize those cues as such. She simply checks to see if the response
fits what her rule proscribes. To this day, she doesn’t get laughter.
She can recognize when it happens, but cannot recognize it for a marker
of the funny, just a marker that something was “funny”. ‘Funny’ is here
not defined in a mentalistic fashion. ‘Funny’ is a placeholder for a
concept to which she has no access. It is an undefined term. What it
means to Professor Grandin is not what it means to people who recognize
something as humorous and then laugh as a result. So, in contrast to
what Currie and Ravenscroft would have us believe, there is no partial
understanding or use of the mental in these cases. There is a pragmatic
rule applied to make sense of the social world. Thus it seems that there
is still a pervasive problem persons with autism face with the mental.
There is also one final point that is troubling for the Simulation
Theory. While this paper takes the authors to be correct in that autism
is probably the result of some low level perceptual deficit, they seem
not to go far enough. They think that there is some specific problem in
the recognition and processing of emotion. According to the results
outlined in Section II, the perceptual deficits that persons with autism
face are of a much more involved than simply emotion perception. Autists
seem to not be able to recognize biological stimuli in general. This
might cause or be concomitant with the deficits in emotion perception,
but it is certainly an important factor of the disorder that must be
taken into account.
Coda
Having shown the limitations
of the previous three theoretical treatments of autism we are now in a
position to briefly sketch a new view of the autistic disorder and
outline some benefits of this view. However, there is one final part to
the general critique this paper presents. As noted above, the main
theoretical treatments of autism, ToM, Executive Control and Simulation
Theory, are seriously troubled. None would predict the variety of
perceptual deficits that persons with autism are now known to have. The
ToM is doubly threatened by these new findings. Again, the perceptual
difficulties require an entire reworking of the modularity the view
relies upon. While this is certainly not an impossible task, such a
reformulation seems rather ad hoc. The Executive Control view
also assumes that the perceptual inputs are veracious. The problem
facing the person with autism is that they are unable to adequately deal
with the variety of elements in their memory. They become confused or
simply mismanage the information. If the social cues were simply not
attended to, then we might see exactly the sorts of behavioral issues
that face persons with autism. They seem perfectly able to problem solve
in certain domains, as recent studies support, they simply do not attend
to the proper events in order to be able to understand the social world.
As with the ToM approach, Executive Control seems develop its
understanding of autism in the wrong way. Finally there is the
Simulation Theory. Simulation Theory, as with Executive Control and ToM,
simply would not predict the variety of perceptual deficits persons with
autism seem to show. While Simulation Theory does place the perception
of basic emotional cues as important, this seems too narrow to account
for the variety of perceptual difficulties facing persons with autism.
Simulation Theory might be more able to handle the additional perceptual
deficits, but it runs into serious troubles in attempting to proffer an
explanation of cases like those of Professor Grandin.
In sum, it seems that these theories are in serious trouble. It is
offere that we attempt to reformulate our view of autism based on the
new research findings. As such the theory can be free of the various
theoretical difficulties facing the prior three theories. This new
theory can also avail itself of research from other disciplines that
have not been involved in the earlier theoretical treatments of autism.
It is to a brief sketch of a new theory of autism that we now turn.
Based on the recent studies on autism, it seems that autism is likely
the result of early perceptual deficits: deficits in attending to social
cues. It is generally thought that autism is a disorder that is present
from birth. It is thought that the initial deficits of autism have a
cascade effect that
hinders the development of later abilities (Gopnik and Meltzoff, 1998).
Much like the disorder PKU
(phenylketonuria) which can cause a sort of mental retardation if gone
unchecked and untreated, autism can prevent the proper cognitive
abilities from developing since the child with autism does not attend to
social stimuli. The recent findings mentioned in section II, also seem
to provide an explanation of some of the behaviors infants and
pre-linguistic children with autism display. These children often avoid
eye contact and become distressed at tickle games or basic parental
hugging and affection. The children seem unconcerned to interact with
others, preferring instead their own pursuits (the word autism comes
from means alone). These behaviors now seem explicable when you take
into account the perceptual deficits that children with autism have.
They simply are never in a position to develop an understanding of the
basic cues and as such never develop the proper cognitive strategies to
understand the behavior of others. While there is still a great deal
more needed in developing this account, the direction seems proper given
the evidence. However, beyond the fact that this view seems to comport
well with new studies, it seems that there are other benefits this new
view can bring.
One benefit that can be taken from taking into account the perceptual
difficulties that persons with autism face is as follows. Over the
years, there has been a general worry that the search for one cause of
autism might be misguided (Ozonoff, et al., 1991). Since autism is
technically a syndrome, it might be that there are multiple causes that
interact to cause the autistic disorder. This sort of idea can be seen
in Ozonoff et al. (1991) where they hypothesize that the same area of
the brain in which is responsible for our ToM abilities is also related
to our executive control abilities. If this is the case then damage to
the ToM area would likely have effects on executive control even though
these are separate abilities. Another benefit comes by way of a partial
solution to a problem facing many researchers when dealing with autism.
One of the difficulties facing researchers when dealing with autism has
been a lack of any definite biological marker of the disorder. However,
as the findings mentioned in section II seem to intimate, there is
progress. And there is evidence from other research on autists as well.
Recently much has been made of the fact that autists have shown
verifiable differences in areas of brain activation (fusiform gyrus)
when perceiving faces (Volkmar, et al., 2003).
The general thought is that the gaze monitoring difficulties that
persons with autism evidence, and the difficulties many persons with
autism claim they have in seeing faces are all linked because such
activities are based in the same neural circuitry that is used in facial
processing. Without going into too much detail on the various areas of
the brain involved, the general point this paper wishes to make is this.
Finding perceptual deficits in modalities outside of vision, like the
ones in audition, may show that the perception of social stimuli
involves a suite of neural mechanisms of which vision is only one. This
would open up the possibility that our understanding of
intentional/social/biological categories is more than modality specific
and requires intermodal mapping of specific inputs in order to come to a
full understanding of social world. Intermodal activity would also
comport well with the Cepenione, et al. findings which demonstrates that
speech is processed later in perception. So whatever processing occurs
regarding “social” stimuli occurs later than pure sensation. If there is
a breakdown on the development or fidelity of this system, we might
plausibly expect effects like those we see in autism.
Researchers working with neuroimaging data, mathematical modeling of
intentional action perception, understanding the role of the “mirror”
neurons in imitation are all undertaking new research in order to answer
the general question of whether there are neural circuits of social
perception. In one article by Ritscher, et al. (2003) researchers are
attempting to address a specific issue: what kind of mathematical model
is better able to model the human ability to reliably perceive
biological motion? Specifically, the researchers are testing whether or
not one mathematical model could produce the correct results or is it
the case that we need multiple independent models working together to
perform the requisite computations? In another article by Puce and
Perrett (2003) they argue that the “mirror” neurons of the STS (superior
temporal sulcus) and the role they seem to play in imitation provides
the beginning of an explanation of part of the mechanism of social
learning. What we see here is a gathering of research addressed at more
basic components of social cognition, especially with an interest to
locate areas and or mechanisms responsible for such abilities. The
findings and conclusions of much of the research attempts to relate some
more basic neuroscientific findings to the various data we have on the
development of ToM. Viewing autism as the result of low level deficits
would comport well with such research.
The
author would like to thank John Bickle, Robert Richardson, and the
reviewers of SJI for their help in the preparation of this manuscript.
Footnotes
ToM knowledge is generally
viewed as innate because a reliance on the Evolutionary
Psychological (EP) argument that claims our ToM knowledge was
selected for and thus made a heritable adaptation (for more see
Barkow, Cosmides and Tooby’s, The Adapted Mind).
Baron-Cohen is specific that his program is to be consonant with
the EP view. ToM theories take it that the young child’s ability
to understand epistemic states in others, that these epistemic
states have causal efficacy, is innately given.
While I believe that
Simulation Theory is likely correct in its application to some
instances of how we predict others’ behavior, it is not the main
tool we use, nor is it the primary one for the development of
our understanding of other minds. Simulation may work in our
more reflective moments when we try to think through a person’s
choice by imagining what we might do given the situation, but it
seems an untoward development of early socio-cognitive
abilities.
Bibliography
Barkow, J.,
Cosmides, L., Tooby, J. (1992). The Adapted Mind. New York.
Oxford University Press.
|
|
