Functionalism providing veridical perception

advertisement
Matthew Menchaca
Two solutions to the interface problem and the qualia theory of perception
The interface problem is that there doesn’t seem to be any clear explanatory
connection between consciousness and the physical order of the world. A solution will
need to explain how consciousness is connected with behavior and/or brain states. This
problem should be contrasted with the easy problem of consciousness. To solve the easy
problem all one needs to do is give a complete account of which brain and/or behavioral
states correlate with each conscious mental state. Functionalism offers one solution to the
interface problem. The solution requires a reanalysis of consciousness. Consciousness is
a complex product of mental states that are to be understood in terms of the functional
role they play in a system. Propositional attitudes are a subclass of contentful states. The
language of thought is another solution to the interface problem. The common way of
speaking about folk belief-desire psychology is relied heavily upon, as are considerations
about the nature of language in general. Consciousness is analyzed as fundamentally
representational where its quality depends upon an underlying computational process
over representations. This is much the same as sentences in a language depend upon an
underlying structure (i.e., a grammar) for their meaning. In this paper I will make these
solutions clear and then I will support functionalism with a theory of perception called
qualia theory.
Section 1: Clarifying the problem
Before explaining how functionalism and the language of thought theory solve the
interface problem more should be said to define the problem. I have said that the interface
problem should be contrasted with the easy problem of consciousness. The reason for this
is that conscious mental states bear no resemblance to brain and/or behavioral states.
Mental states such as “believing that ____” or “thinking that “____” have a structure, but
it is not like the neural structure of the brain, nor like the mechanical structure of
behavior. In addition, both brain states and behavioral states are satisfactorily described
when the physical substrate of those states meets some pre-established criterion of
adequacy. This isn’t the case with mental states: we have no criterion of adequacy for
descriptions (contrary to what the language of thought theorist supposes). Granted, there
are no perfect physical descriptions for neural or behavioral states either. This goes to
show that even the “easy problem” is actually incredibly difficult. Neuroscience and
biology form two disciplines dedicated in part to giving full descriptions of physical
states but these disciplines are not complete.
I have given one indication as to how mental states are picked out: they are picked
out by sentences like “believing that ____” and “thinking that ____” etc. This however
won’t do because sentences, it cannot be assumed, take on the same form as the actual
experiential quality a conscious system has when that system believes or thinks
something. The subject of the interface problem in need of explanation may best be
understood non-symbolically (as in the form of sentences) but as group of non-interpreted
fundamental qualities of experience. I will return to a characterization of that noninterpreted quality in my argument for functionalism in section 4. For clarification at this
point, taking the object of explanation not as the subject of sentences of the form “S
believes that ____,” but rather as the “what it is like for S to believe that ____” is
adequate. Functionalism and the language of thought offer clear ways for this what it’s
like quality to be connected with the physical order of the world.
Section 2: Functionalist solution
1
Matthew Menchaca
Functionalism doesn’t deny that conscious experience involves mental states.
Rather it offers a theoretical model for the causal production of what it’s like qualities
and a method of comparison over systems capable of what it’s like qualities.
Consciousness is understood in the theory as an amalgamation of functional states in a
complex enough organization. Just what kind of complexity is necessary and sufficient
for causing what it’s like qualities is made clearer by analyzing systems as on a
continuum.
Close to one end of this possibly infinite continuum are amoebas, and close to the
other end are systems far more conscious than human, while humans lie somewhere on
the more conscious side. According to the theory to be on the continuum the system must
process information and exhibit some kind of behavior. The input of the functional states
which any of these systems have will be in terms of information, and the output of the
functional states will be in terms of behavior. The kind of information available to each
system on the continuum can be given phenotypically. A complete description is the
object of biology, but specifically those related to cognitive functioning are important for
analyzing the variety of experiences typical to a type of system. For instance one of these
descriptions which may be true of the possible human phenotypical expressions may be
that the perceptual system can only sense a limited spectrum of light frequencies:
ultraviolet frequencies give us no information, while they do for other systems like some
species of fish. Not all information will therefore be available as possible input for a
system. Given this what behavior is possible for a systems development is constrained by
its possible input (this is meant to incorporate the systems genetics as well as the systems
adaptive history within the environment).
It is claimed by this theory that if behavior is observable of a system by another
system, that behavior will be an indication as to the complexity of the functional roles in
place within the observed system. However, it is important to realize that the degree of
complexity over possible input and the kinds of complexity internal to the functional
states necessary for what it’s like qualities will be constrained by the very range of
possible inputs the observing systems can have (as well as its functional character)1.
However, given these problems, the theory offers a first step towards solving the
interface problem. It does this by focusing on systems functions, the degree of specificity
required of the functions in identifying information and the degree of complexity of the
inter-functional relations of the system. These core aspects come together to give a causal
explanation for the production of what it’s like qualities (assuming a fine grained enough
analysis of the content of functional states is possible to give) given a multi-system
comparison of those functional states.
It is assumed by more complex systems that other systems represent, but this is
not necessary for there to be content in these functional states. The content of a functional
state is spread out over the entire input-function-output analysis. Furthermore only some
contents are what it’s like contents. There are more fine grained ways of analyzing the
adaptive nature of functional states within a system, and it will be the intuitive appeal of
those analyses that leads one system to characterize another as capable of some
experiential states and not others.
This just means that humans are limited in their observational capacities and biased, but it is
important to say it in functional terms.
1
2
Matthew Menchaca
An example will clarify the strength of this theoretical model before seeing how
the language of thought offers a different solution to the interface problem:
A frog is a complex informational system, but it is not capable of certain kinds of
what it’s like experiences. Its visual system is capable of identifying movement, which
involves sensation as of an object existing at one time in one place, causing behavior
corresponding to the objects existing at another time and place. Suppose a frogs’ visuals
system is active in just this way and a frog shoots out its tongue to capture a fly. We do
not say of the frog that it is capable of the experience “as of a fly,” even if that is the
object of one of its cognitive capacities. We do not because it just so happens that by
sending another non-fly object of comparable width and height before the frog we
witness its tongue shoot out. What the content exactly is of this specific functional role in
the frog is defined by the possible adaptive behaviors it leads to and the other functional
states which it is connected to. If you are skeptical now on the issue of whether
functionalism solves the interface problem that is okay. I will support functionalisms
solution in section 4. I believe that the qualia theory of perception will supplement the
theory nicely.
Section 3: The language of thought (LOT) solution
The language of thought’s solution to the interface problem I believe is to assume
that the problem is solved. To assume that the problem is solved the LOT theorist takes
as a working hypothesis that mental states are representational. It is just in the nature of
the way we think that our thoughts are manipulations of representations of various states
of affairs. These representations may or may not be of anything in the world though it is
assumed that they are. These representations have a systematic relationship to each other:
there are rules about how they may be formed into a coherent picture of the world and
how an agent may be situated in that world (much like a grammar situates the meaning of
words in a sentence). When in common language one produces the sentence “S believes
that ___” the working hypothesis of the LOT theorist is that S bears some relation (as of
believing) to a string of symbols with representational content equivalent to the
proposition/subject of the belief. So if the ____ is filled with “the cat is on the mat” the
subject is said to be in a believing relation to the-symbol-for-cat in the “on” relation to
the-symbol-for-mat. If the representation is a complex whole then the subject can simply
be said to be in the believing relation to the-cat-on-the-mat. How fine grained may the
symbols of representation be? Infinite, as the representational complexity of an agent
may be theoretically infinite. Of course the conditions for a given representation, i.e., the
place in the world an agent is will constrain this. It is said by the theory that the theory is
implicitly assumed in folk belief-desire psychology language.
How can it be said that this theory assumes that the interface problem is solved?
Remember from section 1, when the problem was explained in detail: it was said then
that an analysis of the special kind of mental states (the experiential kind) in terms of “S
believes that ___” would just amount to the assumption that sentences are the form of
thought a conscious system has when it believes or thinks something. It is exactly an
assumption of this theory that mental states of the form “what it is like for S to believe
that ____” form no special problem. In other words the computational model suggested
by this theory deals with the practical hypothesis forming and hypothesis testing of
agents which are assumed to have the experiential quality. Thus it is assumed that the
problem is solved, or the problem is not a problem.
3
Matthew Menchaca
However, it does not seem like the quality of experience is representational, even
though a particular thought may endorse the use of symbols for its formal expression.
Such thoughts must be systematic and infinitely generative. The kinds of thoughts which
call for the use of this symbolic structure require a theory of perception. It is not my
current object to provide that theory as I am going to for functionalism.
I just want to remark, before I move forward, that what is interesting about these
kinds of thoughts is their application within comparable systems for analyzing
understanding. Since the LOT theorist will be in the enterprise of giving accounts of the
representations necessary for understanding a particular thought, if the functional
mechanisms underlying the production of the representation can be identified, facts
regarding the physical state of a system will advance solutions regarding the easy
problem. This comports well with the common manner in which we speak about the
beliefs and/or desires of others in relation to their understanding. For one, we speak about
individuals representing states of affairs. Secondly we speak of their understanding what
we take to be represented if they can produce comprehensible strings of symbols give
some input (linguistic symbols for instance).2
It is questionable that the theory can apply to systems on a continuum like the
functionalist theory does. For example it seems at least possible that a system might
represent in a language of thought, it may express all the formal requirements involved in
symbol manipulation and yet not have no experiential qualities. I imagine a calculator as
just that kind of system. Of course a calculator expresses no behavior and so it may be
removed from serious consideration, but the point is that the LOT theory takes for
granted that psychological theories range over some predetermined set of systems and
that may be a problem for it. If a problem may be made on this issue then there will be
another reason to demand that the assumption made by the LOT theorist should be given
more justification.
Section 4: The qualia theory of perception
Formally the theory states:
“S has an experience as of a property F iff S senses Fly” and “S senses Fly” is to
be understood to mean “S sense [e-___ and e-___,…, and e-___] where the e___’s are bound in the appropriate way.
“e-___” means “of experience ____ where the ____ can only be filled by
properties of experience (i.e., what it is like to experience). For example the term “esquarely” will be the way the particular squareness of the experience is itself experienced.
The theory can be understood as it denies or accepts one of three principles:
1. Denies the phenomenal principle, which states: If there sensibly appears to a
subject to be something which possesses a particular sensible quality then there is
something of which the subject is aware which does possess that quality.
2. Denies the representational principle, which states: All visual experiences are
representational.
I will not discuss the problems with this interpretation as to how the theory would evaluate
understanding, but only point out that I have been generous to suppose that the symbols in the LOT
should represent the objects that they do. It is also dubious that perception is representational (as I
have remarked before—this will need to be supplement by a theory). Even though it seems like
perceptual experience has to be representational for any one individual to evaluate another’s’
understanding, assuming a sense datum theory of perception may do equally well.
2
4
Matthew Menchaca
3. Accepts the common factor principle, which states: phenomenologically
indiscriminable perceptions, hallucinations, and illusions have an underlying
mental state in common3.
The most important commitment of the theory is its denial of the phenomenal
principle. By denying this principle the theory is able to avoid committing to a dualistic
metaphysics. The acceptance of this principle leads to a dualistic metaphysics in cases of
hallucinatory visual experiences because there must be something of which a subject is
aware that possess whatever qualities the subject is hallucinating, but by postulation it
cannot be a worldly object and be a hallucination, so it must be a mind dependent object.
The second denial is also important. Though it may seem that it is in the very nature
of visual experience that objects are being represented it may be that the representation as
of objects will be part of some rather than other information processing systems and I
believe it is not necessary to add complications where they are not necessary so this is a
unnecessary assumption.
Finally the acceptance of the common factor principle pertains to our discussion in
support of functionalisms solution to the interface problem because functionalist will be
able to identify just what those underlying common mental states may be. In fact
explaining the similarities between the three kinds of visual experience (hallucinatory,
illusory and veridical) in terms of an underlying phenotypically developed functional
mechanism in the organism seems perfectly plausible.
Remember that some content bearing mental states in the functional analysis need not
have content of the what it’s like variety. I will exemplify how the contents of the variety
associated with visual experiences “as of a red tomato” exist for a system given three
variations on input-function-output relations and how they share a common element.
These can be described in turn:
1)
For the veridical experience: the input is of all relevant stimulus required for
the substrate of the functional mechanism to perform its operations (ostensibly
neurons firing are the substrate for the functional mechanism). These
operations conditionally results in the systems interacting with the tomato.
2)
For the illusory experience: the input is of all relevant stimulus required for the
functional mechanism to perform its operations. However, the functional
mechanism is flawed. Within the system this is only apparent conditional upon
the systems adaptation to conditions which result in the flaw4. Outside the
system the functional mechanism responsible may be tested for under a variety
of conditions (which is exactly what is done in illusion experiments).
3)
For the hallucinatory experience: the input is of all relevant stimulus required
for the functional mechanism to perform its operation. For this to be the case
the relevant stimulus for this particular functional mechanism may be
identified as one not directly causally related to the external world (i.e, the
All of these principles are taken from William Fish, “Introduction: Three key principles” Philosophy
of perception: 1-9
4 The fact that illusions exist can only be made sense of, given this analysis, after the fact, when a
species adapts, and thus when the content of its functional states changes and thus when the relevant
stimulus in the environment results in a difference in information.
3
5
Matthew Menchaca
stimulus is in the head)5. Within the system a more abstract procedure wherein
the system relies upon some other of its adapted functional mechanisms will be
necessary to discover at which times a hallucination is present. Those
components of memory I have in mind for human systems. Individuating the
functional mechanisms may best be done causally. I hope that the
interconnectedness of the functional mechanisms is made apparent in
hallucinatory experiences.
I have now shown what functionalism has to add to an endorsement of the common
factor principle which is at the heart of the qualia theory. Given functionalism
insights the common factor principle can be written: In each instance of
phenomenally indistinguishable veridical (perceptual), hallucinatory, and illusory
experience the common mental state that is shared between them for a given system is
the one whose content has all of the relevant stimulus required for that specific
functional mechanism to perform its operation in the system.
Finally it is possible to propose a solution to the interface problem using the
functional theory of mind and the perceptual theory of qualia. (This will be an
analysis of the what it’s like for S to experience that ____.) It is claimed by qualia
theory that to have an experience as of a ____ one is sensing in a particular manner,
and the subject of ones sensing are the properties of experience. Given what is
implied if functionalism is true about the common mental state which holds regards
of the actual state of the world, namely that they share all the relevant stimulus and a
relevant functional mechanism, it is in principle and in practice possible to identify a
correspondence between just the relevant functional mechanism and any property of
experience. If qualia theory is true it will then be possible to comprise the sensing of
experience out of the identifiable properties of experience.
There may be problems with such a theory, most certainly there will be objects,
but that is in the nature of any solution to the interface problem. I will consider
objects and provide replies as I find them voiced by others or as I am able to become
introspectively aware.
This may seem ad hoc, but it is not. Remember that it is postulated that the visual experience is as of
a hallucination and thus we can say truly of the stimulus to the functional mechanism that its cause is
not in the external world. This is at least what hallucinations are currently taken to mean.
5
6
Download