Phenomenology
and Neurophenomenology: An interview
The following is an interview conducted by Michal Sasma for
a Czech journal on May 15, 2003 following my Hebdomades Lectures at the
Philosophy Department of the Univerzita Palackého v Olomouci,
Czech Republic (May 12-15, 2003). This was a series of eight lectures
entitled "Phenomenological interventions in the cognitive
neurosciences" and delivered over the course of a week. Olomouci in the Czech Republic is
located near Husserl's hometown, and is the town where he attended elementary
school. Philosophically, the town
is also known as the place where Wittgenstein spent some time as a young
soldier in the Austrian army.
There's a strange story about him spending time thinking, on the roof of
the city hall, sitting on a couch which he put up there with the help of the
son of the mayor. I don't believe
it, although it is likely that he was there as a soldier. Also, someone suggested that only one
American before me gave the Hebdomades Lectures (all others being Europeans). The American was said to be a guy named
Quine -- I'm not sure that I believe that either, but I don't mind spreading
rumors.
Univerzita Palackého
The
space between phenomenology and the cognitive sciences
Sasma: ... Now let me inquire
further into the idea of neurophenomenology or naturalized phenomenology. In
the introductory essay to the famous collection of papers called Naturalizing
Phenomenology,
the editors (Roy, et al., 1999) had to be very persuasive to show that it is
possible to naturalize Husserlian, transcendental phenomenology. Still, there
are also other, non-transcendental phenomenological approaches that might be more
compatible with the idea of naturalization. Prominent among these is the
body-focused phenomenology of Merleau-Ponty. Now, I wonder precisely which
type of
phenomenology suits best your idea of mutual constraint between phenomenology and
cognitive science? Or, to limit your answer to a pair of alternatives, which do
you prefer, the Husserlian idea of phenomenology as a kind of almost
geometry-like science looking for a priori structures of experience - or
the empirically corroborated phenomenology of Merleau-Ponty which has given up
the transcendental reduction?
Gallagher: Yes, I think I take more
from Merleau-Ponty than from Husserl. And that's one difference between me and Varela,
who, although influenced by Merleau-Ponty, was more influenced by Husserl. I've
always thought that Merleau-Ponty, if he were alive, would be working with the
cognitive scientists - I don't think he'd have any problem with that. The
difficulty Varela was wrestling with is a certain kind of anti-naturalism in
Husserl's thought. Varela was going head-on at phenomenology in the sense that
he took Husserl's phenomenological reduction and tried to make use of it within
experimental science. In contrast, I have a much easier time because I am much
more on the side of Merleau-Ponty, and that also implies the kind of approach I
would take in terms of how you get phenomenology into the empirical sciences
experimenting with cognition, consciousness and so forth. Varela talks about -
talked about - using phenomenological methods, training subjects in the
phenomenological method and using that in the context of experimental science.
Antoine Lutz, who has continued experiments that he started with Varela, has shown
that in certain contexts of neuroscientific practice this actually works (Lutz,
et al., 2002). My own view of how phenomenology could be used in experimental
science presents, I believe, a wider set of possibilities: in addition to
Varela's neurophenomenological approach you can also take insights from
phenomenologists and their analyses and influence the kinds of questions that
are asked in the experiments. In other words, you can influence the
experimental design by basing it on concepts and distinctions that have been
developed in phenomenology. This
is what I have called "front-loaded" phenomenology, because we are
taking phenomenology seriously at the beginning stage of experimental design
(Gallagher, 2003).
Sasma: If I understand you
correctly, then your mission within the cognitive sciences is to make use of
certain concepts or conceptual frameworks developed by phenomenologists, particularly by
Merleau-Ponty, in empirical research. On the other hand, Varela and his
followers are actually trying to put phenomenology in practice, whether by doing it
themselves or by training their experimental subjects. I must confess that the
latter approach seems to me more consequential, in fact more
"phenomenological", than the former, even though I realize it's not
quite clear why it should still be called "philosophy". Anyway, the
first kind of approach seems to be treating phenomenology as a completed and,
in a way, dead sum of knowledge. How about bringing your own phenomenological insights
into play?
Gallagher: I think you can do both. I
don't think phenomenology is dead in any sense. You don't have to take your
phenomenological concepts just from Husserl or Merleau-Ponty, but you can also
take their work and actively extend it and use phenomenological methods to make
distinctions
and to ask different kinds of questions. And then you can take whatever results
you can gain from such analyses and use them in your own scientific
experiments. Or, if you don't get involved directly in the science, you can at
least interface with experimenters and try to influence the kinds of questions
they're asking and the kind of procedures they're using. So I don't think of
phenomenology just as the texts of Husserl and Merleau-Ponty; it's indeed a practice - I agree with Varela on
that. You practice phenomenology, and then you see if it can work to inform
some experiments. Also, it's
important to see where that goes - to see whether the results of the
experiments suggest further phenomenological refinement.
Sasma: So here we have two
different ways in which phenomenology could contribute to cognitive science.
But what about looking at it from the opposite perspective - how precisely can
cognitive science help us make progress in phenomenology - for it's a mutual constraint that's being
talked about, not a one-way constraint?
Gallagher: Actually, "mutual
constraint" is Varela's phrase (Varela, 1996). The phrase I used, in the article that appeared just after
Varela's, is "mutual enlightenment" (Gallagher, 1997). I think “constraint"
might be too strong. Still, I think even this more radical view is defensible.
What are neuroscientists trying to achieve by looking at neurons in the brain?
They are trying to explain some kind of experience. And here comes
phenomenology with its methodical approach to experience saying "take a look, this is what you're trying to
explain, this is what your experiments should be about". Of course, a lot
of cognitive scientists say “we can't really use phenomenology; we don't really
want to pay any attention to that", but in fact they have to because
either formally or informally they have to talk about experience at some point.
Still, the other way - the putative constraint phenomenology receives from
cognitive science - is harder to argue out. Phenomenologists start objecting,
especially those who understand phenomenology as a purely descriptive and
controlled looking at what experience is, which is free from any influence of
theoretical constructs. Because - these phenomenologists would argue - it doesn't
matter how the brain operates, it's not going to change experience;
phenomenological descriptions of consciousness will remain the same no matter
what the neuroscientists or psychologists discover.
Sasma: And what's your own opinion
about this?
Philosophy Courtyard
The
gorilla and phenomenology
Gallagher: I think it's possible that
the science can certainly inform phenomenology. Cognitive science is able to discover
things, even about experience, that are out of the phenomenologists' reach. I'm
in a permanent search for good examples of this, but let me try one out here.
There are studies of change-blindness, where the strategy is as follows: a
person is looking at something and she is instructed to pay close attention and
to describe some aspect of what she is seeing. In her concentrated attention she somehow fails to notice a
very obvious change that takes place in what she is observing. Not a change in her conscious
experience, but a change in the object attended to. An excellent example of this is on the web - an
experiment by Daniel
Simons at the University of Illinois - where subjects are shown a
30-second video in which some people kick around a ball. And the instruction
you give to your subject is this: "Take a look at this video. What I want
you to do is count the number of times the ball changes hands". Now, as the subject is looking, a guy
dressed in a gorilla suit walks into the middle of this group of people and he
stops, looks directly into the camera, waves his hands wildly and then he walks
off. And then you finish the experiment by asking "OK, so how many times
did the ball change hands?" I
did this little experiment with my students in class. You get students raising their hands and saying "I
counted 37 times" and another student raises her hand: "No, no, no,
it was 36 times." So you have a discussion of how many times the ball
really changed hands and so far nobody has mentioned the gorilla. And then you
ask "OK, and how many people saw the gorilla?" And they'll say
"What are you talking about? What gorilla?" And they are all very
surprised when you play the video again and they see a gorilla. In fact they suspect you have fooled
them: "How could it be the same video? I didn't see the gorilla there
before!" Now, this is a wonderful experiment about what we don't see if
we're focused on something, if our attention is caught up in something. So,
what would a phenomenological analysis of this case be? Or more precisely, what
exactly should the phenomenological description of that be? If I'm counting the times
the ball changes hands and in my visual field, right in front of my nose a
gorilla enters and I don't notice it, what should a phenomenologist say about
that? Well, here's an experiment that tells us something about our experience.
I think it's a good example of how cognitive science could inform
phenomenology.
Another
example I've been thinking about - but I'm not sure if it works - comes from
Dennett who mentions a very simple experiment: I ask the phenomenologist to
describe what is in his peripheral visual field. Now it depends on how good this phenomenologist is. Most people if asked what they can
experience in their visual field will provide a relatively detailed
description. They tend to say that
everything in the peripheral field looks relatively clear. In fact, though -
and you can perform this test on yourself - if I fix my eyes at the focal
center of my visual field, and a large and bright red object is introduced into
my peripheral vision, I will have no idea what color that object is. Our
peripheral vision is not as sharp as we think it is.
Slunecko: This might be due to the
fact that the important information a living organism can get from its
peripheral visual field - for instance in case of being endangered - is
connected with movement.
Gallagher: Yes, that's right, you get
movement from your peripheral field much better than you get color or
shape. I'm not sure how
phenomenologists would do in this kind of experiment. We would expect them to bracket out the automatic answer
given by most people about the acuity of their peripheral field. So the
experiment might not tell them more than they already know. Indeed, as I think of it, this might be
a case of what we should call experimental phenomenology - and something that
phenomenology can tell the cognitive sciences.
Sasma: I have to confess that I'm
not totally convinced by these examples. Both of them do show that statistically most subjects ignore or
respectively misinterpret the way their visual field is structured or
influenced by attention. This, however, doesn't rule out the possibility that
some subjects - call them "phenomenologists", "good
disinterested observers" or whatever - could be trained, either by
themselves or by the experimenters, to attend to their experience in a substantially more subtle way than their
untrained colleagues, as you have just suggested in regard to the second
example. (This point is being argued out by some scholars writing about usage
of vipassana meditation in cognitive science.) Talking about the first example, it's
likely, or at least possible, that such trained subjects would be able to come
up with descriptions that would provide some account of the change of the visual
field caused by the gorilla-man - even though this account wouldn't need to
amount to the subject's explicitly reporting the gorilla-show. The other
example is different from the first one in that all it shows - if I understand
it correctly - is that the subjects in question have a wrong theory of perception, not that there
are limits to the perceptual experience itself which only science can provide
to it "from the outside", so to say.[1]
And again, the trained subjects may do one hundred percent better here. Let me
be still more specific and suggest one possible way the trained subject could
respond to
the gorilla test, if taken in a phenomenological fashion. What would his
description of the experience be? I think it would have to contain a note of the
kind "I was concentrating just on the ball and I didn't really see what
else there was on the screen". Now, if you add a note like that to an
otherwise naive description, you will get a sort of alibi for the first-person
phenomenological account against the third-person charge of not seeing the
gorilla-man. The careful description would be the first-person account of the
experience while the gorilla itself would belong to the third-person account.
Gallagher: That's true. I would say that
a good phenomenological account of my experience when I am there trying to
count the number of ball exchanges would not include the gorilla. Because I'm
concentrating, I didn't see the gorilla there. So the phenomenological
description would leave the gorilla out. Objectively, third-person, you would
say the gorilla is there though - but that itself is an interesting fact! Not
only for the cognitive scientist but for the phenomenologist as well. It tells
the phenomenologist that his description of a visual field might leave
something out that is objectively there. This "something" is not in
my experience, and then the question is "why isn't it in my
experience?" Indeed, the
phenomenologist might even claim that he has given a good account of what was in
front of him; and certainly he would be interested to know that he missed the
gorilla that was in front of him.
Sasma: It might be represented somehow in the experience, but not
necessarily as the gorilla...
Gallagher: If it is, that's even
better! It might be represented somehow, as an "absence" or however you want
to put it, but if it is, does the phenomenology capture it as such? And the
scientist will say it should capture it. Well, cognitive science in this sense is
at least enlightening the phenomenologist about something that has to do with
with the structure of attention - and this is surely something the
phenomenologist should be interested in.
Sasma: So far we have been thinking
of the relationship between cognitive science and phenomenology either very
generally or on a rather terminological level (Should we speak about a
"constraint" or "enlightment"?). Let's go more to the heart
of the matter now: Is the cognitive scientist actually in a position to correct the phenomenologist's
descriptions? Is he entitled to tell the phenomenologist "you've missed
something, so try to look better next time" etc.?
Gallagher: If the cognitive scientist
sends the phenomenologist back to the drawing board, so to speak, thinking
"I'd better review the phenomenological description to see what's really
there" or something like that, I think it's too late. The phenomenologist
has already given his description. If you then tell the phenomenologist
"You know, there was a gorilla there. Go back and see what you can find"
- that's already changing the experience. The phenomenologist can't go back to
the original experience - he is now going to be looking for a gorilla, and that
just changes everything. Rather,
the lesson cognitive science teaches the phenomenologist is that phenomenology
has certain limitations. You could describe in some sense your experience and
yet, there might be something there you're missing - and I don't mean the
gorilla, but something about the nature of experience. And this might just be an implicit part
of experience in some fashion. It might be a structure or indivisible part of
your experience that you are leaving something out when focusing. That might be
an important fact for phenomenology. It tells us, the cognitive scientist as
well as the phenomenologist, something about the nature of attention, something
about what attention does to our visual field.
Sasma: But if phenomenological
description leaves in principle something out, what about the difference I have
mentioned between truly phenomenological description made by a trained or so to speak
"professional" phenomenologists on one hand and
"phenomenological" description made by just any subject on the other?
Can we still maintain that difference as a difference in principle - and what should it be based
on?
Gallagher: OK, I am tempted to say
almost immediately, that's an empirical question. We'd have to have an
experiment: have a non-phenomenologist from the street do the gorilla
experiment and have the phenomenologist sit next to him and do the gorilla
experiment and provide a phenomenological description and then see what the
difference between them is. It's difficult because on the one hand the
instruction was not to give a phenomenological description of what you see -
the instruction was to count the number of times the ball changes hands. And
that's a different kind of task. How would the phenomenologist go about giving
a phenomenological description of what happens when he actually addresses that
task and starts counting the ball exchanges? Should he be paying attention to
counting the exchanges? But he would also have to provide some kind of
phenomenological description of what that experience was like. Things get very
complicated by other questions. How precisely is phenomenology different from
what the instructed subject might be doing? Could someone count the exchanges
under a phenomenological reduction? I think a phenomenologist would say yes.
And then still, it's a difficult task to give a phenomenological description of
that. So I think we really have a nest of problems here that would be difficult
to solve. And I'm not sure if I am going to be able to give you an answer now
that would satisfy anybody on that question.
Sasma: Could you specify what kinds
of sentences are representative for phenomenology as the method? We will
probably agree that sentences like "There were 36 ball-exchanges" are
not prototypical phenomenological descriptions.
Gallagher: That's right. The
phenomenologist is not really interested in counting the ball exchanges. He is
interested in the experience of counting the ball exchanges. So he is interested
more in structural features of what's happening when he is engaged in that kind
of operation.
Sasma: But to do that, to unveil
all the structural features of his experience while fulfilling a task, the
phenomenologist has to be using the scientific account; he has to be using the
fact that
there is something missing in his direct experience, right?
Gallagher: No, I don't think that the
constraint the cognitive scientist brings to the phenomenologist is "Oh,
your description is wrong." The description is not wrong - it's the
description of the experience and maybe the phenomenologist would then be very
concerned to identify structural aspects of that experience. But I don't think that the scientist is
then in any kind of position to say, "There, you are wrong about that
description - that's not what you experienced". I think that's a possible objection of the phenomenologist:
"No matter what the cognitive scientist tells me, he's not going to be
able to tell me that I'm wrong about my experience". And I think that's fine.
Another way for the cognitive scientist to put it is to say that there is
something missing: "Your phenomenological description is not wrong, it has
just left something out that was there" - and that should be interesting
enough for the phenomenologist. That should give phenomenologists some pause
and motivate them to say, "Yeah, now what does that mean in terms of the experience
that I had?" What is it about
my experience, or about the structure of my experience, that prevents me from
seeing the gorilla. Somehow the
experience was rather focused, so it tells me something about attention.
Nothing the cognitive scientist will say is going to change the phenomenological
description, but in some sense the experiment will point out the limitations of
that description - and the fact that there's something going on that you can't
capture with a simple phenomenological description might be very important,
especially in thinking about consciousness.
Sasma: But even if the message of
the cognitive scientist to the phenomenologist stays with a note that his
description is missing something, then for the phenomenologist to get interested, the
"missing" object must be relevant to his own business. In other
words, both of the guys have to share a common object of reference. Otherwise
the phenomenologist could always avoid the direct interface by arguing:
"The cognitive scientist is pointing at objects that I have no ambition to
catch in my description - the real objects of the world. I've bracketed these out in the epoche and represent them in my
description as the purely formal correlate 'x' of my intentional acts".
The phenomenologist will always be justified in responding "so what"
to anything the cognitive scientist points out.
Gallagher: Well, most phenomenologists
do say "so what." And the idea is "well, maybe we shouldn't say
'so what'." Maybe we should say 'let's figure this out' or 'let's think
about this'. I think that there is
a common object of reference. Both
the phenomenologist and the enlightened cognitive scientist are interested in
the experience. I want to get the
cognitive scientists to say that too - they are interested, they need to be
interested in the experience.
Obviously, the experimenters are not interested in explaining the
gorilla - the object in the real world - they are interested in explaining why,
in our experience, we seem to miss the gorilla.
Slunecko: Back to the problem of
common frame of reference. When the phenomenologist says "In my
phenomenological description there's no gorilla," the cognitive scientist
could hypnotize him and demonstrate to him that he's got the gorilla in his
memory of the experience - he's attended to it unconsciously. So the phenomenologist's
description is - should be - about the gorilla, the problem is he just doesn't
know.
Gallagher: That would be a very
interesting complication. If you could be hypnotized and suddenly in your experience you could recall seeing a
gorilla - why didn't that register in the first place? Why didn't it show up in
my phenomenological description? That's a good point. Beyond that, I have
enough difficulty talking to phenomenologists about non-conscious perception -
a concept that is very well developed in neuropsychology, but is thought to be
a contradiction in phenomenology.
Sasma: So let us leave this rather
detailed and fragmented discussion for a while and sum up the main points. What
are the hardcore ideas of your project of cooperation between the two
disciplines - what's your final verdict about their "mutual
enlightment"?
Gallagher: So, phenomenology-cognitive
science. Well, as we've said it's not that the cognitive sciences are supposed
to improve
the phenomenology but somehow rather, in my word, enlighten it - to make the
phenomenologist think about some things differently. A lot of this, I think,
depends on how you define phenomenology. To give you a definition, I will have
to go back to when you were asking me: "Is it Husserl or
Merleau-Ponty?" - and I said "It's Merleau-Ponty". As I've said,
unlike Varela I'm thinking more in line with Merleau-Ponty, and maybe I even
borrow some of the insights of the early Heidegger. This is not to say I'm
leaving Husserl out, especially the later Husserl (I mean his genetic
phenomenology - but even the early Husserl has a lot of resources to offer, as
for example, Jean-Luc Petit has pointed out in regard to Husserl's early
discussions of kinaesthesia - especially interesting in light of the recent
work on mirror neurons). Merleau-Ponty himself read Husserl, of course. He went
to the Husserl archives in Leuven to study his texts, and was inspired
especially by Ideen II. But then Merleau-Ponty was also very interested in reading psychology
and neurology - he was interested in looking at case studies (for instance the
famous case of Schneider in Phenomenology of Perception), and he wanted to carry
phenomenological insights into the interpretations of those case studies,
trying to show how phenomenology has something to offer psychology and
neuroscience. So today, Merleau-Ponty would be, I think, right in there saying
"this is enlightening my phenomenological discourse, my understanding of
what experience is." So if we define phenomenology as an attempt to
understand what experience is, then we can certainly use both, the
phenomenological method as outlined by early Husserl (as long as we understand
his anti-naturalism in the right framework, and not as anti-science), as well
as phenomenology as understood by Merleau-Ponty. Remember Merleau-Ponty's comments on the phenomenological
reduction in Phenomenology of Perception - that it's incomplete or really incompletable
- and then you'll see that it opens the door of phenomenology a bit to the
possibility that it can learn something, maybe even learn from the cognitive
sciences about its own limitations.
Philosophy
Department
Phenomenology
and heterophenomenology
Sasma: When you spoke about the difficulties
your idea of "mutual enlightment" encounters both among scientists as
well as phenomenologists, you mentioned that a large number of cognitive
scientists (and, I must add, some philosophers of mind coming from the
analytical tradition) have a hard time granting phenomenology the important
role we believe it in fact has in formalizing the explanandum, i.e., experience. This
negative attitude, it strikes me, might be a result of a curious widespread
ignorance of the technical, methodological meaning the term has in Husserl's
founding texts. Instead, "phenomenology" is often simply used as a
synonym of "subjectivity" - not as a formalized account of experience
but as that experience itself.
Gallagher: Yes, scholars use the term
“phenomenology" in different ways. Even psychologists use the word
“phenomenology" quite frequently as another word for experience, not for a
method. So, of course, we have to keep in mind that the usage of the word
depends on the kind of tradition we are part of.
Sasma: A good example of an emptied, flattened usage of the
word is, I think, Dennett's much disputed "heterophenomenology". I suppose that as
phenomenologists in the strong sense of the word, we would agree that Dennett's
project of heterophenomenology doesn't seriously harm our position and his
arguments against what he calls "first-person science of
consciousness" are somehow tricky. Perhaps we all vaguely feel that
Dennett's behaviorist skepticism, so to say, is unable to grasp the full
meaning of experience, it mistreats the phenomenological evidence. Is there a
healthy grain in our intuition or can Dennett mock us for being mere
folk-psychologists?
Gallagher: Well, first of all, what Dennett calls first-person
data generated in experimental situations is already in a sense predetermined
by the kind of tasks, questions, and the instructions the experimenter gives to
the subjects. And quite frequently what happens is that the experimenter will
ask very specific questions because they're interested in specific kinds of
issues and data. That, in a sense, is already introducing theoretical
considerations into the generation of that phenomenological data. The
first-person data is already shaped, pre-figured and pre-determined by the kind
of questions that are being asked by the experimenter. So, first of all, what
we're starting with are just reports of experience and certainly not the methodological,
phenomenological descriptions. So we're taking reports of experience that have
been shaped already by the experimenter and then Dennett is taking and treating
those simply as third-person data (as bits of narratives or whatever), and
measuring up that first-person data with those other third-person data sets
that are generated in the experiment. So, what that means basically is that you
cannot say that Dennett is treating first-person data seriously and certainly
what he is starting with is totally different from what phenomenologists start
with.
Sasma: Husserl's idea of
phenomenology rests heavily upon a supposition that there are some a priori structures of our experience
(be it the noematic apriori of eiditic invariants or noetic correlational apriori). Can one still think in
terms of a priori structures of consciousness in the context of front-loaded (or neuro-)
phenomenology? Wouldn't it often be the case that where the phenomenologist
could see "a priori connections" between certain aspects of experience, the
neurologist would find just accidental groupings of neurons? Could you name a
couple of examples of such survivors of a priori structures in naturalized
phenomenology? (the temporal frame? the Gestalts?...)
Gallagher:
It depends on how one defines a priori structures. There is a neo-Kantian idea that
proposes a relative apriori -- not as absolute as Kant's apriori. Also, there is a tradition
in theoretical biology (e.g., von Uexkull) that identifies certain biological
capacities (shaped by evolution) as apriori for the organism. So to the
extent that neuroscientists would talk about hard wiring and certain limits
placed upon perception or cognition by that hard wiring, then one could
identify a priori constraints / conditions. So relatively stable structures of conscious
experience (although disruptable by brain damage, etc.) might count as apriori. More abstractly, one might
say that a statement like "All experience is embodied experience"
might represent a kind of transcendental fact. I would think that the latter is
not something that would change at all, and is something of an absolute
(although obviously not accepted by everyone). Some other things could change
over time or by circumstance, but not that.
Sasma:
This non-Kantian definition sounds sympathetic. Applied to our case, a
priori structures
of experience are those that are relatively stable (and not necessarily
"independent from experience", as Kant would have it) within the life
of an organism. What needs to be added, I think, in order for phenomenology to
do useful work, is the condition that at least some of these structures must be
graspable by primarily first-person accounts. It is this second condition that
Dennett disputes. According to him, the only reliable account of experience is
the one provided by behaviorist-like analysis of the interviewing scientist.
What would you respond to this?
Gallagher:
I would ask Dennett how he would define rationality -- and therefore, how to
define science. Certainly, Dennett doesn't want to give up science. So I would
throw the issue back to him. How can he justify the idea of science without
some way of accounting for first-person experience -- at least the first-person
experience of the scientist that somehow allows the scientist to do science?
Fragments
from the history of AI: from computationalism to the enactive approach.
Petru: The era of cognitive science
was initiated by the outburst of cybernetics and information theory in 1940's
and 1950's, and until very recently it was governed by the metaphor of the
brain as a kind of computer. What is left of this idea of mind as a
computation?
Gallagher: Well, I don't think there's
a lot left. I think it's usually now considered to be a wrong starting point.
We've started with the computational model and then tried to explain the brain
with it. In 1990's we made great progress in understanding how little we know
about the brain. So now that we are getting a little bit of knowledge about it,
we also are starting to realize that it's a much better approach to start with
the brain and to learn from it as much as we can, and then, with that knowledge
at hand, to go back into the computational fields and try to build neural
networks in connectionist models (and dynamic-systems models) that would do
more justice to the question of our brain. So if the whole task of AI is to
build something like intelligent machines then what we've learned, I think, is
that it's better to go in that other direction, starting with mapping our
brains and trying to build something like an artificial brain. The problem with
computationalism is that the brain is not simply a computational machine, it's
much different and we are still learning how it works.
Sasma: In their book Embodied
Mind,
Francisco Varela and his co-authors subdivided the history of cognitive science
into three phases: computationalist era, connectionist era, and finally the
present era of what the authors call "enactive," open-ended approach.
What profits do you think could this last approach bring to the aim you've
mentioned - creating an artificial brain? And further, how do you see the
chances of the enactive approach in trying to influence or even to enter the
mainstream philosophy of mind, AI and cognitive science?
Gallagher: That's a big question. The
way I understand the enactive, open-ended idea, is that it's not enough just to
think about what's going on in the brain. It wouldn't be enough to build an
artificial neural system. The
brain is a part of the body, and the body, which is located in an environment,
also contributes to experience. So we have to re-define what we call system. The old definition of system
was "whatever is in the box", e.g., whatever is in this computer chip
or whatever is in the brain - "here's the system that we're trying to
explain". So we tried computationalism, and it failed; now we are trying
to build the system from the ground level according to the principles that we
know from the study of the brain, but still, we don't have the complete story.
The complete story has to put the brain into the body and the body into the
environment. And then we define the system as all of those elements added
together. And we should add that this system is extended in time (it takes time
for experience to develop), and that the environment is also social. I don't think that we can explain
experience, cognition, or even the brain, without taking intersubjectivity into
consideration. And so what we have
to understand are the interactions that occur between the organism and the environment, between one
organism and another, and within the organism between the brain and the rest of
the body (Gallagher, 2004). So that's a much bigger job than the older
conceptions of cognitive science. But I think that if you ignore these other
larger parts of the system, you're never going to get the smaller parts of the
system right. Because it is a system. So if we just try to design artificial brains
without taking into account the embodiment or ecological aspects of our
experience in the environment, then we're going to fail. That's what I think
Varela is suggesting, and Merleau-Ponty too, and I think they were right.
Sasma: The enactive theory would
certainly satisfy any philosopher, but how do scientists - according to your
experience - react to it? Do they take it seriously?
Gallagher: I think that some scientists
might say that it makes sense at a kind of ideological level. For some it is in
the distant background. Others,
like Gibsonian ecological psychologists, and Varela (remember that he was a
scientist first, and to the last), may want to move it forward. But in terms of a particular scientific
project, one has to figure out ways to test it. In this sense it can operate as
a background theory, as an emerging paradigm, and it can work in a way similar
to the way that a paradigm guides the interpretation of observations and
experiments. Scientists think in this way: "Right now I am trying to
understand how Broca's area contributes to language" - they ask even more
specific questions in order to get their experiments off the ground.
Furthermore, experimentation requires you to control elements of the system
that you are not testing. So one has to ignore the bigger picture in the actual
experimental practice. Science is
done as one small project at a time, and it is often difficult to see the
bigger picture in the small projects.
One always has to take a step back to see the bigger picture. I think that's where philosophy comes
in.
References
Gallagher, S. (In press, 2004). How the
Body Shapes the Mind. Oxford: Oxford University Press.
Gallagher, S. 2003. "Phenomenology and
experimental design," Journal of Consciousness Studies 10 (9-10): 85-99.
Gallagher, S. 2002.
"Hyleticka zkusenost a prozivane telo" Trans. Michal Sasma. Philosophica (The Czech Republic) 5: 103-126; translation
of "Hyletic Experience and the Lived Body," Husserl Studies 3 (1986): 131-166.
Gallagher,
S. 1997. Mutual enlightenment:
Recent phenomenology in cognitive science. Journal of Consciousness Studies 4 (3): 195-214.
Gallagher, S. and A. Meltzoff.
1996. "The Earliest Sense of
Self and Others: Merleau-Ponty and Recent Developmental Studies," Philosophical
Psychology 9: 213-236.
Gallagher, S. and J. Cole.
1995. "Body Schema and Body
Image in a Deafferented Subject," Journal of Mind and Behavior 16: 369-390.
Gallagher, S. 1986. "Body
Image and Body Schema: A Conceptual Clarification," Journal of Mind and
Behavior 7: 541-554.
Lutz, A., Lachaux, J.-P., Martinerie, J., and
Varela, F. J. (2002), Guiding the study of brain dynamics using first-person
data: Synchrony patterns correlate with on-going conscious states during a
simple visual task. Proceedings of the National Academy of Science USA: 99,
1586-1591.
Roy, J.M., Petitot, J., Pachoud, B., and Varela, F.
(1999), Beyond the gap. An introduction to Naturalizing Phenomenology. In J.
Petitot, F. Varela, B. Pachoud and J.M. Roy (ed). Naturalizing Phenomenology, (pp. 1-80).
Stanford, CA: Stanford University Press.
Varela, F. (1996), Neurophenomenology : A
Methodological Remedy to the Hard Problem. Journal of Consciousness Studies, 3: 330-50.
[1] Michal Sasma: To prove this, one would have to argue that certain kinds of phenomenological descriptions necessarily contain mistakes which cannot be corrected “from within" (from the 1-st-person phenomenological perspective). Nevertheless, to prove that CS could enlight phenomenology amounts to merely showing that in some specific cases, the CS is faster in identifying the mistakes which however can in principle be reecognized as such also “from within".