October 2, 1998
1st Annual Lecture
Daniel Dennett
Co-Director, Center for Cognitive Studies
University Professor
Austin B. Fletcher Professor of Philosophy
Tufts University
Note: A portion of the Q&A has been removed from the transcription at the request of
Daniel Dennett. The transcription has also been edited for clarity by Professor Dennett.
Things About Things
Lecture Introduction
Moderator
It gives me the greatest pleasure to welcome you here today to the first annual Benjamin
and Anne Pinkel lecture on Mind-Brain Paradigms.
I would like first to take this occasion to thank Sheila Pinkel, on behalf of the students
and faculty, of the University of Pennsylvania for her generous gift on behalf of the estate
of her parents, which has made this annual series of lectures possible.
I can think of no more fitting tribute to the memory of Benjamin Pinkel than the creation
of a forum for the continuing discussion and investigation of the fundamental questions
concerning the nature of the mind, which were his intellectual passion.
Mr. Pinkel, who received his BSE in Electrical Engineering from the University of
Pennsylvania in 1930, sought in his monograph "Consciousness Matter and Energy" to,
and here I quote Mr. Pinkel, "propose an expansion of the scientific view of nature to
include a concept of mind." In light of this proposal it seems especially appropriate that
these lectures should be sponsored by our university's Institute for Research in Cognitive
Science. This Institute, which is home to the only National Science Foundation Science
and Technology Center devoted to the study of cognition, has as its mission, the
development of a scientific understanding of cognitive processes, and the creation of
technologies based on this understanding - a mission very much in harmony with Mr.
Pinkel's vision.
I am delighted to introduce Sheila Pinkel, who will share with us her special perspective
on this vision.
Sheila Pinkel
Thank you very much, it's an honor to be here today, and on behalf of introducing the
first lecture in this series on Mind-Brain Paradigms, I thought I would tell you a little
about my family.
My father graduated from the Moore School of Electrical Engineering at the University
of Pennsylvania in 1930, and then began working at NACA, the National Advisory
Committee on Aeronautics, which was later to become NASA, the National Aeronautics
and Space Administration, first at Langley Field and then at the Louis Laboratory in
Cleveland Ohio. My mother graduated from William and Mary at the top of her class
and immediately joined the editorial staff of NACA at Langley Field where she started
editing my father's reports. They soon married, and moved from Virginia to Cleveland
Ohio. During World War II, dad devoted his energies to perfecting the cooling system
for the jet engine, joining the legions of people fighting to develop technology to win the
war. He went on to design the first nuclear reactor at the Louis Laboratory, headed one
of the largest divisions of scientists at the Lab, and spent a great deal of his professional
life working on nuclear airplane propulsion systems. In 1956 he and the family moved to
Santa Monica so that he could join the Rand Corporation, to function as a consultant to
the Air Force on novel airplane propulsion feasibility studies.
When he retired in 1972 he began to focus full-time on his passion and interest, the
philosophy of mind. He took numerous courses in neuroscience at UCLA, and
voraciously read neurological research and mind-brain philosophy, in an attempt to
understand how the neurological system, including the brain worked. As he continued
this research he became fascinated by the remarkableness of the neurological system, and
increasingly believed that conventional descriptions of brain functioning could not
account for the phenomenon of mind, which he viewed at a kind of energy in nature
which had not yet been accounted for by physicists. When he investigated what people
called the physical world, he found in fact that matter is described as a miniscule nucleus
amidst a huge area vaguely described as a "field" containing rings of electrons. What that
field is made of, or how it asserts its energy, is not understood, so that while we call
matter material, in fact there is almost nothing solid about it, and what that energy is,
which holds the atom together, is also not understood. We do have the word "field,"
however, and in his view, this word and words like it are used in science to give the false
impression that the physical world is well understood. In fact, from his point of view we
do not understand magnetism, gravity, the strong force or the weak force, the four forces
identified as the fundamental forces in nature. In addition, a force which is not
acknowledged in this list is the force of mind. He did not believe in psychokinesis; he
did approach the demonstration of energy of mind as a scientist would look for proof.
What one must do is to think, "pick up this pencil," then mind instructs the body, and
one's hand picks up a pencil. So he believed that, while we cannot explain energy of
mind any more than we can explain the other forces, we should not ignore it, but rather
include it in the list of forces in nature. In fact this was a very radical way of
restructuring our understanding of the physical and mental world. What impressed me
most in my discussions with him was his open attitude about knowing, and an ability to
imagine an alternative to the prevailing structure. During the last years of his life every
time I visited him he would describe yet another feature of the neurological system,
which is awesome in it constructive process. It struck me that what he had done was to
give perspective to human understanding of physical and mental structures, and it was in
acknowledging the remarkableness that he found the unimaginable. Ultimately I came to
understand that he was really asking, "of what benefit is it to believe that it is just a
matter of time until human being understand it all? Doesn't it serve us better to believe
that there is far more in this world than we can possibly imagine, and that by opening up
to the possibility of the unimaginable we can open up to new paradigms and new
solutions that we cannot now imagine." My family and I are quite grateful to the
University of Pennsylvania for giving us the opportunity to sponsor the current series on
Mind-Brain Paradigms. In 10 years we hope to publish a book of these lectures and
discussions, as a way of making the dialogue on this subject visible. The speakers
participating in this series will expand and extend the ideas which were so fascinating to
my father. My parents would have been proud to participate in this series, and I look
forward to watching this dialogue unfold.
Thank you again.
Moderator: Thank you very much Sheila.
Speaker Introduction
We are most fortunate to have Professor Daniel Dennett here today, to inaugurate this
series on Mind-Brain Paradigms. Professor Dennett, who is Distinguished Arts and
Sciences Professor, and Director of the Center for Cognitive Studies at Tufts University,
is an internationally-renowned scholar in the philosophy of mind and cognitive science.
He is the author of numerous books and essays, which have profoundly influenced
thought about the mind and its relation to nature over the past three decades. His works
include Content and Consciousness, which appeared in 1969, followed by Brainstorms,
ElbowRoom, The Intentional Stance, Consciousness Explained, Darwin's Dangerous
Idea, Kinds of Minds, and most recently, Brainchildren.
Today's lecture will be followed by a panel discussion; we are very fortunate to have our
distinguished colleagues Professor Gary Hatfield, recently chair of the philosophy
department, and Professor Robert Seyfarth, currently chair of the University's psychology
department, as discussants for today's lecture.
And without further ado, I am delighted to present Professor Dennett, to lecture on
"Things about Things."
Lecture : Things About Things
Daniel Dennett
It's a very great honor to me to be invited to give the inaugural lecture in this series. 1 I
am delighted to be here in this beautifully refurbished room. I want to thank Mrs. Pinkel
for setting this wonderful program up, and for the copy of the book by her father, which I
will take home with me and treasure. It's a particular honor to be invited as a philosopher
to give this lecture, because philosophers aren't always in such high regard in scientific
quarters. In a review of Steven Pinker's book, How the Mind Works, in New York Review
of Books, the British geneticist Steve Jones had the following comment to make: "To
most wearers of white coats, philosophy is to science as pornography is to sex. It is
cheaper, easier and some people seem, bafflingly, to prefer it." Now that view is all too
common, and I understand it from the depths of my soul. I appreciate why people think
this, but I think it is also important to combat this stereotype in a friendly and
constructive spirit, and no place better than in a center for research in cognitive science.
What philosophers can be good at – there aren't many things we can be good at – is
helping people figure out what the right questions are. When people ask me whether
there's been any progress in philosophy I say, "Oh yes, mathematics, astronomy, physics,
physiology, psychology – these all started out as philosophy, and once we philosophers
got them whipped into shape we set them off on their own to be sciences. We figured out
how to ask the right the questions, and then we turned them over to other specialists to
answer.
Of those fields that have been born out of philosophy, perhaps the most recent is
psychology, the study of the mind, and some people would say it was a premature birth; it
should have been kept in the oven a little longer. This is why it is such a rich field, I
think, for philosophers coming to cognitive science these days, and it is so delightful to
find people in white coats struggling with our issues, grappling with the questions that we
philosophers have been grappling with for a few thousand years. They have come to
realize these questions are not that easy. And one of the great side benefits of the boom
in works on consciousness by neuroscientists and physicists and psychologists that we've
seen in the last decade, is that writing these books has been a humbling experience for the
authors. It is hard to know when you're asking the right sorts of questions, and whenever
that is your problem, you’re stuck doing philosophy. And so it is gratifying to me that
the Institute for Research in Cognitive Science has declared an interest in having a
philosopher come and give the inaugural lecture in this series. On behalf of my
discipline, I am delighted to be here.
Perhaps we can all agree that in order for intelligent activity to be produced by embodied
nervous systems, those nervous systems have to have things in them that are about other
things in the following minimal sense: there is information about these other things not
just present but usable by the nervous system in its modulation of behavior. (There is
information about the climatic history of a tree in its growth rings--the information is
present, but not usable by the tree.) The disagreements set in when we start trying to
characterize what these things-about-things are - are they “just” competences or
dispositions embodied somehow (e.g., in connectionist networks) in the brain, or are they
more properly mental representations, such as sentences in a language of thought,
images, icons, maps, or other data structures? And if they are “symbols”, how are they
“grounded”? What, more specifically, is the analysis of the aboutness that these things
must have? Is it genuine intentionality or mere as if intentionality? These oft-debated
questions are, I think, the wrong questions to be concentrating on at this time, even if, “in
the end”, they make sense and deserve answers. These questions have thrived in the
distorting context provided by two ubiquitous idealizing assumptions that we should try
setting aside: an assumption about how to capture content and an assumption about how
to isolate the vehicles of content from the “outside” world.
A Thing about Redheads
The first is the assumption that any such aboutness can be (and perhaps must be) captured
in terms of propositions, or intensions - sometimes called concepts. What would an
alternative claim be? Consider an old example of mine:
Suppose, for instance, that Pat says that Mike “has a thing about redheads.” what Pat
means, roughly, is that Mike has a stereotype of a redhead which is rather derogatory and
which influences Mike’s expectations about and interactions with redheads. It’s not just
that he’s prejudiced against redheads, but that he has a rather idiosyncratic and particular
thing about redheads. And Pat might be right - more right than he knew! It could turn out
that Mike does have a thing, a bit of cognitive machinery, that is about redheads in the
sense that it systematically comes into play whenever the topic is redheads or a redhead,
and that adjusts various parameters of the cognitive machinery, making flattering
hypotheses about redheads less likely to be entertained, or confirmed, making relatively
aggressive behavior vis-à-vis redheads closer to implementation than otherwise it would
be, and so forth. Such a thing about redheads could be very complex in its operation or
quite simple, and in either case its role could elude characterization in the format:
Mike believes that: (x)(x is a redhead e . . . . )
no matter how deviously we piled on the exclusion clauses, qualifiers, probability
operations, and other explicit adjusters of content. The contribution of Mike’s thing about
redheads could be perfectly determinate and also undeniably contentful and yet no
linguification of it could be more than a mnemonic label for its role. In such a case we
could say, as there is often reason to do, that various beliefs are implicit in the system.
(“Beyond Belief,” [in The Intentional Stance, p148]
But if we do insist on recasting our description of the content in terms of implicit beliefs,
this actually masks the functional structure of the things that are doing the work, and
hence invites us to ask the wrong questions about how they work. Suppose we could
“capture the content” of such a component by perfecting the expression of some
sentence-implicitly-endorsed (and whether or not this might be “possible in principle,” it
is typically not remotely feasible). Still, our imagined triumph would not get us one step
closer to understanding how the component accomplished this. After all, our model for
such an activity is the interpretation of data structures in computer programs, and the
effect of such user-friendly interpretations (“this is how you tell the computer to treat
what follows as a comment, not an instruction to be obeyed”) is that they direct the
user/interpreter’s attention away from the grubby details of performance by providing a
somewhat distorted (and hyped up) sense of what the computer “understands”. Computer
programmers know enough not to devote labor to rendering the intentional interpretations
of their products “precise” because they appreciate that these are mnemonic labels, not
specifications of content that can be used the way a chemist uses formulae to describe
molecules. By missing this trick, philosophers have created fantasy worlds of
propositional activities marshaled to accomplish reference, recognition, expectationgeneration, and so forth. What is somewhat odd is that these same philosophers have also
largely ignored the areas of Artificial Intelligence that actually do take such content
specifications seriously: the GOFAI worlds of expert systems, inference engines, and the
techniques of resolution theorem-proving and the like. If you want to look at a model of
Mentalese, or the language of thought, look at Prolog, look at some expert systems, look
at the data structures therein. But no philosophers seem to take these seriously as models
of what go on in minds. Presumably they can see at a glance that whatever these
researchers are doing, their products are not remotely likely to serve as realistic models of
cognitive processes in living minds. But then why do they take the idea of a language of
thought seriously if they're not prepared to look at those models? This is a good
unanswered question for those philosophers.
A thing-about-redheads is not an axiomatized redhead-theory grafted into a large data
base. We do not yet know how much can be done by a host of things-about-things of this
ilk because we have not yet studied them directly, except in very simple models - such as
the insectoid subsumption architectures of Rodney Brook and his colleagues. One of the
chief theoretical interests of Brooks’ Cog project is that it is pushing these profoundly
non-propositional models of contentful structures into territory that is recognizable as
human psychology. Let’s see how they work, how they interact, and how much work
they can do before we take on the task of linguifying their competences as a set of
propositions-believed.
I want to continue my harping on the theme of philosophers playing a role, because it's
important to realize that the direction that Rod Brooks is now famously taking is a
direction that was argued for many years ago by a philosopher, and he earned a great deal
of hooting derision for his efforts. That, of course is Hubert Dreyfus, who claimed way
back in 1972 that in order to be intelligent you have to have a body. The artificial
intelligence community rose up en masse and said he was crazy. When I talk about this
before AI audiences, I use an overhead that says "Just because Bert Dreyfuss said it,
doesn't mean it couldn't be true." Now people have come around to seeing that maybe
Bert was right about something, even if he put it in suspiciously aprioristic and
philosophical terms.
Cog has become something of a media star, and has been featured in so many television
documentaries on robotics and artificial intelligence that I hardly need introduce it to you.
[At the lecture I showed some video clips, from which the following indented text is
drawn, introducing Cog, and the Cynthia Brezeal’s “emotional infant” robot, Kismet:
In order to act intelligently, there's a lot of things you have to know about the world. And
one approach is to try and tell an artificial intelligence program, write it out in great detail
and tell it all the facts. By building a robot, we're trying to build a system which can act
in the world, interact with people, and learn for itself. Our hope is that that will lead to a
quicker accumulation of the sort of knowledge of what it is to act in the world, so that we
can have true artificial intelligence. To encourage people to interact with the robot
naturally, we've built the robot to look like a human and to act like a human. He has two
eyes, microphones for ears, and a set of gyroscopes to give it a sense of balance. Each of
Cog's eyes has two cameras, one that has a very wide angle, peripheral field of view, and
one that has a very narrow field of view, but much higher resolution. Cog has a total of
21 degrees of freedom, including two 60-degree of freedom arms, three degrees of
freedom in the torso, three in the neck, and three in the eyes. . . .
...This is Kismet, Kismet is my infant robot, it gives me facial expressions which tells me
what its motivational state is. This one is anger, [...] disgust, excitement, fear, happiness,
this one is interest, this one is sadness, surprise, this one is tired, and this one is sleep. In a
suitable learning environment, Kismet's drives are in homeostatic balance. This means
that the robot is neither understimulated, nor overwhelmed by its interaction with the
caretaker. Stimulation intensity is computed by the perceptual system, moving faces are
a social stimuli whose intensity is proportional to the amount of motion. Any other
motion is viewed as a non-social stimulus. Kismet works with the caretaker to keep the
perceptual stimuli within an acceptable range. Kismet's emotions and expressions reflect
its motivational state. By reading Kismet's facial expressions, the caretaker can respond
to the robot's needs and stimulate the robot appropriately. One of Kismet's drives is to be
social. If Kismet does not receive any social stimulation, it becomes lonely and looks
sad. The caretaker responds by making face-to-face contact with the robot. This satiates
the social drive, and Kismet displays happiness. However, if the social stimulus is too
intense, Kismet becomes asocial and shows disgust. This is a cue for the caretaker to
back and restore the interaction to a suitable intensity level.]
The opponent process, homeostatic system for these quasi-pseudo emotional states is
actually much more subtle than is suggested in that bit of video. Once it's ported over
onto Cog itself, it will play a big role in allowing Cog to get its infant education from
many different human interactors. The idea is that Cog is going to go through a period of
infancy, and is going to learn a lot about the world the way you and I do, by playing with
things, reaching for things, learning about occlusion and gravity and bumping things and
discovering things that it can't get that it wants, and having people around it all the time.
If you're going to make this work, you have to make the robot as engaging as possible to
people, and the team has already succeeded beyond the predictions of many of the
skeptics. People faced with Cog or Kismet often make fools of themselves the same way
people do when encountering a darling baby in a baby carriage. The sense they have that
there is an agent, a self in there, whose interests they begin to care about very deeply, is
already very potent. This confirms a point I've been making to philosophers for years.
Some philosophers have said that if you ever did make a conscious robot you'd have a
problem, a civil rights problem, convincing the world that it was conscious, and not just a
zombie. And I've said, no, it's actually going to turn out to be the other way around.
Long before you have a conscious robot, you're going to have pseudo-conscious robots
and the hard thing is going to be convincing the world that they aren’t conscious. We're
already beginning to see this, but I have no idea what percentage of those who encounter
Cog come away with the conviction that they have been in the presence of another
conscious being.
Transducers, Effectors, and Media
The second ubiquitous assumption is that we can think of a nervous system as an
information network tied to the realities of the body at various restricted places:
transducer or input nodes and effector or output nodes. In a computer, there is a neat
boundary between the "outside" world and the information channels. A computer can
have internal transducers too, such as a temperature transducer that informs it when it is
getting too hot, or a transducer that warns it of irregularities in its power supply, but these
count as input devices since they extract information from the (internal) environment and
put it into the common medium of information-processing. It would be theoretically tidy
if we could identify the same segregation of information channels from "outside" events
in a body with a nervous system, so that all interactions happened at identifiable
transducers and effectors. The division of labor this permits is often very illuminating. In
modern machines it is often possible to isolate the control system from the system that is
controlled, so that control systems can be readily interchanged with no loss of function.
The familiar remote controllers of electronic appliances are obvious examples, and so are
electronic ignition systems (replacing the old mechanical linkages) and other computerchip-based devices in automobiles. And up to a point, the same freedom from particular
media is a feature of animal nervous systems, whose parts can be quite clearly segregated
into the peripheral transducers and effectors, and the intervening transmission pathways,
which are all in the common medium of impulse trains in the axons of neurons.
At millions of points, the control system has to interface with the bodily parts being
controlled, as well as with the environmental events that must be detected for control to
be well-informed. In order to detect light, you need something photosensitive, something
that will respond swiftly and reliably to photons, amplifying their sub-atomic arrival into
larger-scale events that can trigger still further events. In order to identify and disable an
antigen, for instance, you need an antibody that has the right chemical composition.
Nothing else will do the job. It would be theoretically neat if we could segregate these
points of crucial contact with the physics and chemistry of bodies, thereby leaving the
rest of the control system, the "information-processing proper," to be embodied in
whatever medium you like. After all, the power of information theory (and automata
theory) is that they are entirely neutral about the media in which the information is
carried, processed, stored. You can make computer signals out of anything - electrons or
photons or slips of paper being passed among thousands of people in ballrooms. The very
same algorithm or program can be executed in these vastly different media, and achieve
the very same effects, if hooked up at the edges to the right equipment.
As I say, it would be theoretically elegant if we could carry out (even if only in our
imagination) a complete segregation. In theory, every information-processing system is
tied at both ends, you might say, to transducers and effectors whose physical composition
is forced by the jobs that have to be done by them, but in between, everything is
accomplished by medium-neutral processes. In theory, we could declare that what a mind
is is just the control system of a body, and if we then declared the transducers and
effectors to be just outside the mind proper--to be part of the body, instead--we could
crisply declare that a mind can in principle be out of anything, anything at all that had the
requisite speed and reliability of information-handling. Now this important theoretical
idea is close to being the Grand Enabling Assumption of cognitive science. It has
liberated theorists for more than two decades from having to cope with the unimaginable
complexities of neural connectivity and interactivity.
This important theoretical idea sometimes leads to serious confusions, however. The
most seductive confusion is what I call the myth of double transduction: first the nervous
system transduces light, sound, temperature, and so forth into neural signals (trains of
impulses in nerve fibers) and second, in some special central place, it transduces these
trains of impulses into some other medium, the medium of consciousness! This is, in
effect, what Descartes thought, and he declared the pineal gland, right in the center of the
brain, to be the locus of that second transduction. While nobody today takes Descartes'
model of the second transduction seriously, the idea that such a second transduction must
somewhere occur (however distributed in the brain's inscrutable corridors) is still a
powerfully attractive, and powerfully distorting, subliminal idea. After all (one is tempted
to argue) the neuronal impulse trains in the visual pathways for seeing something green,
or red, are practically indistinguishable from the neuronal impulse trains in the auditory
pathways for hearing the sound of a trumpet, or a voice. These are mere transmission
events, it seems, that need to be "decoded" into their respective visual and auditory
events, in much the way a television set transduces some of the electromagnetic radiation
it receives into sounds and some into pictures. How could it not be the case that these
silent, colorless events are transduced into the bright, noisy world of conscious
phenomenology? This rhetorical question invites us to endorse the myth of double
transduction in one form or another, but we must decline the invitation. As is so often the
case, the secret to breaking the spell of an ancient puzzle is to take a rhetorical question,
like this one, and decide to answer it. How could it not be the case? That is what we must
see. I can't answer all of that question today. After all I am a philosopher; I ask the
questions I don't answer them. But I can perhaps make a little progress.
What is the literal truth in the case of the control systems for ships, automobiles, oil
refineries and other complex human artifacts doesn't stand up so well when we try to
apply it to animals, not because minds, unlike other control systems, have to be made of
particular materials in order to generate that special aura or buzz or whatever, but because
minds have to interface with historically pre-existing control systems. Minds evolved as
new, faster control systems in creatures that were already lavishly equipped with highly
distributed control systems (such as their hormonal systems), so their minds had to be
built on top of, and in deep collaboration with, these earlier systems. 2
This distribution of responsibility throughout the body, this interpenetration of old and
new media, makes the imagined segregation more misleading than useful. But still one
can appreciate its allure. It has been tempting to argue that the observed dependencies on
particular chemicals, and particular physical structures, are just historical accidents, part
of an evolutionary legacy that might have been otherwise. True cognitive science (it has
been claimed) ought to ignore these historical particularities and analyze the fundamental
logical structure of the information-processing operations executed, independent of the
hardware.
The Walking Encyclopedia
This chain of reasoning led to the creation of a curious intellectual artifact, or family of
artifacts, that I call The Walking Encyclopedia. In America, almost every schoolyard has
one student picked out by his classmates as the Walking Encyclopedia--the scholarly
little fellow who knows it all, who answers all the teacher's question, who can be counted
on to know the capital cities of all the countries of the world, the periodic table of
chemical elements, the dates of all the Kings of France, and the scores of all the World
Cup matches played during the last decade. His head is packed full of facts, which he can
call up at a moment's notice to amaze or annoy his companions. Although admired by
some, the Walking Encyclopedia is sometimes seen to be curiously misusing the gifts he
was born with. I want to take this bit of folkloric wisdom and put it to a slightly different
use: to poke fun at a vision of how a mind works.
According to this vision, a person, a living human body, is composed of a
collection of transducers and effectors intervening between a mind and the world. A
mind, then, is the control system of a vessel called a body; the mind is material--this is
not dualism, in spite of what some of its ideological foes have declared--but its material
details may be safely ignored, except at the interfaces--the overcoat of transducers and
effectors. Here is a picture of the Walking Encyclopedia.
[figure 1 about here]
In this picture--there are many variations--we see that just inboard of the transducers are
the perceptual analysis boxes that accept their input, and yield their output to what Jerry
Fodor has called the "central arena of belief-fixation" (The Modularity of Mind, 1983).
Just inboard of the effectors are the action-directing systems, which get their input from
the planning department(s), interacting with the encyclopedia proper, the storehouse of
world knowledge, via the central arena of belief-fixation. This crucial part of the system,
which we might call the thinker, or perhaps the cognition chamber, updates, tends,
searches, and - in general - exploits and manages the encyclopedia. Logic is the module
that governs the thinker's activities, and Noam Chomsky's LAD, the Language
Acquisition Device, with its Lexicon by its side, serves as a special purpose, somewhat
insulated module for language entry and exit.
This is the generic vision of traditional cognitive science; For several decades,
controversy has raged about the right way to draw the connecting boxes that compose the
flow charts- - the "boxology" - but little attention has been devoted to the overcoat. That
is not to say that perception, for instance, was ignored - far from it. But people who were
concerned with the optics of vision, or the acoustics of audition, or the physics of the
muscles that control the eye, or the vocal tract, were seen as working on the periphery of
cognitive science. Moreover, those who concerned themselves with the physics or
chemistry of the activities of the central nervous system were seen to be analogous to
electrical engineers (as contrasted with computer scientists).
We must not let this caricature get out of hand. Boxologists have typically been quite
careful to insist that the interacting boxes in such flow diagrams are not supposed to be
anatomically distinct subregions of the brain, separate organs or tissues “dedicated” (as
one says in computer science) to the tasks inscribed in the boxes, but rather a sort of
logical decomposition of the task into its fundamental components, which could then be
executed by “virtual machines” whose neuroanatomical identification could be as
inscrutable and gerrymandered as you like - just as the subroutines that compose a
complex software application have no reserved home in the computer’s hardware but get
shunted around by the operating system as circumstances dictate.
The motivation for this vision is not hard to find. Most computer scientists don’t really
have to know anything much about electricity or silicon; they can concentrate on the
higher, more abstract software levels of design. It takes both kinds of experts to build a
computer: the concrete details of the hardware are best left to those who needn’t concern
themselves with algorithms or higher level virtual machines, while voltages and heatdispersion are ignorable by the software types. It would be elegant, as I said, if this
division of labor worked in cognitive science as well as it does in computer science, and a
version of it does have an important role to play in our efforts to reverse-engineer the
human mind, but the fundamental insight has been misapplied. It is not that we have yet
to find the right boxology; it is that this whole vision of what the proper functioning parts
of the mind are is wrong. The right questions to ask are not:
How does the Thinker organize its search strategies?
or
Isn't the Lexicon really a part of the World Knowledge storehouse?
or
Do facts about the background have to pass through Belief Fixation in order to influence
Planning, or is there a more direct route from World Knowledge?
These questions, and their kin, tend to ignore the all-important question of how
subsystems could come into existence, and be maintained, in the highly idiosyncratic
environment of a mammalian brain. They tend to presuppose that the brain is constructed
of functional subsystems that are themselves designed to perform in just such an
organization - an organization roughly like that of a firm, with a clear chain of command
and reporting, and each sub-unit with a clear job description. We human beings do indeed
often construct such artificial systems - virtual machines - in our own minds, but the way
they come to be implemented in the brain is not how the brain itself came to be
organized. The right questions to ask are about how else we might conceptualize the
proper parts of a person.
Evolution embodies information in every part of every organism. A whale's baleen
embodies information about the food it eats, and the liquid medium in which it finds its
food. A bird's wing embodies information about the medium in which it does its work. A
chameleon's skin, more dramatically, carries information about its current environment.
An animal's viscera and hormonal systems embody a great deal of information about the
world in which its ancestors have lived. This information doesn't have to be copied into
the brain at all. It doesn't have to be "represented" in "data structures" in the nervous
system. It can be exploited by the nervous system, however, which is designed to rely on,
or exploit, the information in the hormonal systems just as it is designed to rely on, or
exploit, the information embodied in the limbs and eyes. So there is wisdom, particularly
about preferences, embodied in the rest of the body. By using the old bodily systems as a
sort of sounding board, or reactive audience, or critic, the central nervous system can be
guided - sometimes nudged, sometimes slammed - into wise policies. Put it to the vote of
the body, in effect.
Let us consider briefly just one aspect of how the body can contribute to the wise
governance of a mind without its contribution being a data structure or a premise or a rule
of grammar or a principle, in a phenomenon modeled by Kismet. When young children
first encounter the world, their capacity for attending is problematic. They alternate
between attention-capture - a state of being transfixed by some object of attention from
which they are unable to deflect their attention until externally distracted by some more
powerful and enticing signal - and wandering attention, attention skipping about too
freely, too readily distracted. These contrasting modes are the effects of imbalances
between two opponent processes, roughly captured under the headings of boredom and
interest. These emotional states - or proto-emotional states, in the infant - play a heavy
role in protecting the infant's cognitive systems from debilitating mismatches: when
confronted with a problem of pattern-recognition that is just too difficult, given the
current immature state of the system, boredom ensues, and the infant turns off, as we say.
Or turns away, in random search of a task more commensurate with the current state of
its epistemically hungry specialists. When a nice fit is discovered, interest or enthusiasm
changes the balance, focussing attention and excluding, temporarily, the distractors. 3
I suppose this sort of meta-control might in theory have been accomplished by some
centralized executive monitor of system-match and system-mismatch, but in fact, it
seems to be accomplished as a byproduct of more ancient, and more visceral, reactions to
frustration. The moral of this story may not strike one as news until one reflects that
nobody in traditional Artificial Intelligence or cognitive science would ever have
suggested that it be important to build a capacity for boredom or enthusiasm into the
control structure of an artificially intelligent agent.4 We are now beginning to see, in
many different ways, how crippled a mind can be without a full complement of emotional
susceptibilities. 5
Things that go Bump in the Head
But let me make the point in a deeper and more general context. We have just seen an
example of an important type of phenomenon: the elevation of a byproduct of an existing
process into a functioning component of a more sophisticated process. This is one of the
royal roads of evolution. 6 The traditional engineering perspective on all the supposed
subsystems of the mind - the modules and other boxes - has been to suppose that their
intercommunications (when they talk to each other in one way or another) were not
noisy. That is, although there was plenty of designed intercommunication, there was no
leakage. The models never supposed that one box might have imposed on it the ruckus
caused by a nearby activity in another box. By this tidy assumption, all such models
forego a tremendously important source of raw material for both learning and
development. Or to put it in a slogan, such over-designed systems sweep away all
opportunities for opportunism. What has heretofore been mere noise can be turned, on
occasion, into signal. But if there is no noise - if the insulation between the chambers is
too perfect - this can never happen. A good design principle to pursue, then, if you are
trying to design a system that can improve itself indefinitely, is to equip all processes, at
all levels, with "extraneous" byproducts. Let them make noises, cast shadows, or exude
strange odors into the neighborhood; these broadcast effects willy-nilly carry information
about the processes occurring inside. In nature, these broadcast byproducts come as a
matter of course, and have to be positively shielded when they create too many problems;
in the world of computer simulations, however, they are traditionally shunned - and
would have to be willfully added as gratuitous excess effects, according to the common
wisdom. But they provide the only sources of raw material for shaping into novel
functionality.
It has been recognized for some time that randomness has its uses. For instance, sheer
random noise can be useful in preventing the premature equilibrium of dynamical
systems - it keeps them jiggling away, wandering instead of settling, until some better
state can be found. This has become a common theme in discussions of these hot topics,
but my point is somewhat different: My point is not that systems should make random
noise--though this does have its uses, as just noted - but that systems should have squeaky
joints, in effect, wherever there is a pattern of meaningful activity. The noise is not
random from that system's point of view, but also not useful to it. A neighboring system
may learn to "overhear" these activities, however, thereby exploiting it, turning into new
functionality what had heretofore been noise.
This design desideratum highlights a shortcoming in most cognitive models: the absence
of such noise. In a real hotel, the fact that the guests in one room can overhear the
conversations in an adjacent room is a problem that requires substantial investment (in
soundproofing) to overcome. In a virtual hotel, just the opposite is true: Nobody will ever
overhear anything from an “adjacent” phenomenon unless this is specifically provided for
(a substantial investment). There is even a generic name for what must be provided:
“collision detection”. In the real world, collisions are automatically “detected”; when
things impinge on each other they engage in multifarious interaction without any further
ado; in virtual worlds, all such interactions have to be provided for, and most cognitive
models thriftily leave these out - a false economy that is only now beginning to be
recognized.
Efficient, effective evolution depends on having an abundant supply of raw material
available to shape into new functional structures. This raw material has to come from
somewhere, and either has paid for itself in earlier economies, or is a coincidental
accompaniment of features that have paid for themselves up till then. Once one elevates
this requirement to the importance it deserves, the task of designing (or reverse
engineering) intelligent minds takes on a new dimension, a historical, opportunistic
dimension. This is just one aspect of the importance of maintaining an evolutionary
perspective on all questions about the design of a mind. After all, our minds had to evolve
from simpler minds, and this brute historical fact puts some important constraints on what
to look for in our own designs. Moreover, since learning in the individual must be, at
bottom, an evolutionary processes conducted on a different spatio-temporal scale, the
same moral should be heeded by anybody trying to model the sorts of learning that go
beyond the sort of parameter-tuning that is exhibited by self-training neural nets whose
input and output nodes have significances assigned outside the model.
Conclusions
Cognitive science, like any other science, cannot proceed efficiently without large
helpings of oversimplification, but the choices that have more or less defined the field are
now beginning to look like false friends. I have tried to suggest some ways in which
several of the traditional enabling assumptions of cognitive science - assumptions about
which idealized (over-) simplifications will let us get on with the research - has sent us on
wild goose chases. The “content capture” assumption has promoted the mis-motivated
goal at explicit expression of content in lieu of the better goal of explicit models of
functions that are only indirectly describable by content-labels. The “isolated vehicles”
assumption has enabled the creation of many models, but these models have tended to be
too “quiet,” too clean for their own good. If we set these assumptions aside, we will have
to take on others, for the world of cognition is too complicated to study in all its
embodied particularity. There are good new candidates, however, for simple things about
things now on offer. Let’s give them a ride and see where we get. Thank you very much
for your attention.
Panel
Gary Hatfield
I chose to go first and respond to Dan Dennett's philosophical talk in the normal
philosophical way, that is, to read a shorter philosophical paper in response to his paper.
Cognitive science arose through an interdisciplinary federation of approaches to mind
and cognition, comprising psychology, philosophy, linguistics, computer science and AI,
and sometimes neuroscience, biological studies of animal behavior, and anthropology.
At its origin, it carried the stamp of three of these fields: linguistics, artificial intelligence,
and philosophy, especially a philosophy of mind closely allied with philosophy of
language. The originating ideology was expressed Jerry Fodor, who treated cognition as
essentially linguistic, and modeled all cognitive processes as transitions among sentences
in the head. Sentences expressed in an innate language of thought, which he compared to
the machine language of a digital computer. Although cognitive science itself has moved
on to a healthy disunity of approaches, variously embracing all of the disciplines I have
named, the core literature is still dominated by the founding ideology. Dan Dennett has
been a participant in the development of cognitive science for three decades, from before
it was known as cognitive science. He has from the beginning been an admirer of AI, but
a critic of the "sentences in the head" view of cognition. He has encouraged the use of
mentalistic language from within the intentional stance, a useful but perhaps provisional
view that treats organisms and artifacts such as heating systems as rational agents. He has
also been a staunch critic of the attribution of determinate phenomenal state to perceivers,
such as in the case of visual perception, perceptual images filled with color and form. He
sees such attributions as falling prey to traditional mentalism, going back to Descartes.
His criticisms in this domain come in conflict with an area of scientific psychology, the
experimental study of perception, which typically does ascribe determinate phenomenal
states, including images containing color and form to perceivers. Finally, Dennett has
urged cognitive scientists, who are usually dismissive toward BF Skinner and other
behaviorists, not to throw the baby out with the bathwater, but instead to preserve useful
parts of behaviorism, especially its learning theory.
Now today Dennett has offered a diagnosis of the current state of the core literature in
cognitive science by questioning two assumptions that still have currency. The first is the
assumption, not as widespread as it once was, that the content of mental states must be
fully capturable without remainder or significant distortion by equating that content with
sentences or propositions. He argues that some content is too amorphous to be rendered
with the precision of English or some other language. Instead, he encourages us to
pursue the tack exemplified by the worker Rodney Brooks, who has built insect-like
creatures that find their way about in complex environments without the benefit of
internal prepositional structures. I think this is good advice. Not that no cognition is
linguistic, or that linguifying models of some cognitive process shouldn't be pursued. It's
just that it would be good to explore other avenues of explaining contentful processes,
and not to assume from the start that all cognitive processes in both humans and animals
must be conceived as sentences in the language of thought or as implicit propositions.
The second assumption is the tendency to treat the nervous system, or the internal
information system, as tied to the body only at specific locales, called transducers and
effectors, which are usually equated with sensory transducers such as rods and cones in
the eye, and motor effectors such as the nerve terminals that control muscle activity.
Dennett complains that this assumption tends to treat cognitive processes as isolated and
insulated from the body in two ways. First, it treats cognitive processes in isolation from
the bodily structures they control. But, he rightly reasons, some or much information
about the environment, information which must be taken into account in explaining the
cognitive and motor achievements of organisms, may be embodied in the structure of the
muscles and the limbs, that internal cognitive processes are assumed to control. Whether
these limbs be leg, wing, fin, or flipper. Here Dennett takes a commendable step in the
direction of a more ecologically-informed cognitive science, which understands that
organisms, their bodies and their psychologies, evolve in relation to environments.
Second, Dennett argues that traditional models regard internal processes of information
transmission as isolated from the brain structures that realize those processes. Which is
to say two different things: that details about brain structure can be ignored, on this view
Dennett criticizes, and that these ignorable brain processes are insulated from other brain
processes. But, he rightly asserts, surely the brain structures that mediate cognition have
evolved have evolved from earlier simpler neural structures. Whatever we can learn of
the history of their evolving function, whether through comparative work or
paleontology, is likely to help in understanding their present function. Moreover,
although brain structures are specialized for the tasks they perform, Dennett suggests that
they nevertheless are likely to be subject to influence or perturbations from the other
structures, and that this might have good effects, of the sort that he mentions in the
connectionist models that don't settle down to local maxima, but he considers these good
effects to perhaps be found on a larger system level.
Dennett's warning about this second assumption of insulation and isolation is a helpful
counter to some models in AI and cognitive pychology, that posit and internal boxology,
conceived independently of ecological constrainsts, and constructed on the assumption
that hardware, or the brain's wetware, doesn't matter. It could also serve as a useful
corrective to some aspects of Fodor 1983 Modularity of Mind book, namely the
conception of modules as fully insulated from other modules, cognitively and presumably
mechanically. There are however two points in Dennett's discussion of the second
assumption that I would qualify. First, where he says that questions such as "Isn't the
lexicon really a part of the world knowledge storehouse?" are not the right questions to
ask, I would say rather that they're not the only questions to ask. As I said at the outset, I
take the increasing disunity of cognitive science to be a healthy thing. It makes sense to
pursue multiple research strategies simultaneously, rather than putting all the eggs in the
basket of a single research school. This division of labor strategy has been defended in
philosophy of science by Phillip Kitcher and Miriam Solomon among others. Second,
Dennett's warning about examining isolated systems could mistakenly be taken as an
indictment of one of the leading research strategies of post-behavioristic psychology.
Behaviorism, especially in its Skinnerian form, encouraged thinking of the organism in
terms of inputs and outputs, stimulus and response, and in solving an equation for what
mediates between them. The ideal was that psychological science should explain the
behavior of the whole organisms, by postulating appropriate laws mediating between S
and R. Although cognitive science rejected behaviorism, the founding ideology pursued
the same ideal, of solving for the behavior of the whole organism, now inserting a
complicated boxology between S and R. This tendency was especially clear in the work
of philosophers who spoke of attributing beliefs and desires to explain the pattern of
external behavior. That goal was incorporated into Fodor's early statements of the
language of thought thesis, and it informed Dennett's intentional stance. Since the demise
of behaviorism, another research strategy has driven the main areas of experimental
psychology. The strategy has been to give up, for now at least, the claim to solve the
organism, and to focus on functioning subsystems within the organism or specific classes
of phenomena. The subsystems include the various perceptual systems, whether visual,
auditory, olfactory or internal to the body, the control of sequential motor action, the
perceptual processes in reading and in listening to speech, and the memory systems. The
phenomena include various dimensions of visual experience and the play of attention. In
these cases, psychologists have fruitfully investigated a single psychological capacity, or
a group of related capacities. They have then sought to explain that capacity, rather than
the behavior of the whole organism, through a functional decomposition of its workings.
A functional decomposition looks something like a subpart of Dennett's boxology, but it
is more narrowly focused that the explanation of the walking encyclopedia. It is likely to
aim at capacities of the organism that are biologically salient, such as distance or color
perception, and it takes into account what is known about the physiology or the
evolutionary history of the structure subserving that capacity. The success of this
research strategy in psychology reveals an important moral which I think is left out of
Dennett's account. The psychologist can study and seek to explain various sensory,
cognitive, and motor capacities even if physiological or evolutionary knowledge is not
available. Psychology can make its own approach to the mind, independent of
neuroscience or evolution. Indeed, this is a good thing, since it helps to know what an
organism can do, and how it does it, if one is to approach our massively complex brains
or our sketchily known evolutionary history, and ask how these capacities are realized
physiologically or how they evolved. (So I'm suggesting there that psychology can and
in fact does often lead the way toward neuroscientific investigations or towards posing
evolutionary questions.)
So far I've commended much of what Dennett says, with a few qualifications. Now I
come to a point of disagreement. At one point, Dennett describes what he a terms a
confusion embodied in the myth of double transduction. According to the myth, "first the
nervous system transduces light sound temperature and so forth into neural signals, and
second in some special central place, it transduces these trains of impulses into some
other medium, the medium of consciousness." He compares this position to that of
Descartes, and suggests that while nobody holds it explicitly, it has a powerfully
subliminal effect on research. Now, it's true that only a very few modern psychologists
and neuroscientists explicitly endorse Cartesian dualism (there have been a few but not
very many) according to which the mind is a special, separate, immaterial substance,
entirely distinct from matter and sharing none of its essential properties. But
contemporary scientists do hold something that is formally equivalent to the doctrine of
double transduction, that is, some contemporary scientists do. I say that it is formally
equivalent to suggest that while the set of relations named in Dennett's doctrine is
preserved, and so its mathematical form, the metaphysics is left aside, so let me explain
this.
The formally equivalent law is found in the part of scientific psychology known as
psychophysics. The name psychophysics describes its original subject matter.
Psychophysics studies the lawful relations between physical stimuli and psychological
experience, between in the case of vision, say, wavelengths of light and the experience of
color they cause in the observer. Psychophysics, and the study of auditory and visual
perception more generally, were the heartland of psychology during its rapid expansion
in the last part of the 19th century. The findings in psychophysics were what for many
people convinced that psychology could be a science. Study of perceptual experience
was and remains a royal road to empirical study of mind, and of the mind-brain. Reports
of phenomenal experience provide access to the products of psychological processes and
to the existence of certain central brain states. Now from its inception, psychophysics
was intended to deal with the physiological side of the relation between stimulus and
experience. That is, it aimed also to discover facts about the neurophysiological events
that yield experience. About 15 years ago two respected psychophysicists, Veneeta
Teller at the University of Washington and Ed Pew here at Penn, formulated what they
took to be an important goal for psychophysicists, one that lay implicit in the science.
They argued for the importance of formulating what they called "psychophysical linking
propositions." A linking proposition makes explicit the reasoning that there must be a
physiological locus, whether in a small area of the brain or across areas, at which
physiological activity is related to the experienced content, what philosophers call
"qualia" and some psychophysicists such as Pew of "sensory experience." Now the
general form of a psychophysical linking proposition is like this: one conceives of a
stimulus, which is mapped by a certain relation onto a physiological state, which is
mapped onto other physiological states, leading finally to a physiological state that still
maps onto other physiological states, but which has the characteristic of being what Pew
and Teller the bridge locus, which is the physiological state that directly maps onto the
experienced content, say a visual, auditory or olfactory perception.
[Explaining diagram] So S here is a stimulus, M are various mappings, phis are all
physiological states, and psi is an experienced content, psychological state and M-star is
the mapping between the physiological state and phi-m is the bridge locus.
Although Pew and Teller do not explicitly call such propositions "laws", it is natural to
read their general scheme, this general scheme here, as giving the form for hypothetical
laws relating brain states or brain activity to qualia. Pew and Teller would then be seen
as proposing that psychophysicists formulate and test various law statements relating
brain states to sensory experiences, to go along with known or newly discovered laws
relating stimulus to experience. In this way, one can narrow in on central brain processes
from both ends, through stimulus pathway (the physiological side) and through the
phenomenal effect. By the way, neither Pew nor Teller are dualists, nor need they be;
they are both materialists. We can now see that what Dennett portrays as an undesirable
implicit assumption is in fact an explicitly formulated and defended tenet of current
science. Since Dennett is usually respectful of science, and usually prefers to cast his
philosophical points politely as recommendation rather than as summary executions of
scientific hypotheses, we need an explanation of why he would reject, outright, the
legitimacy of a proposal like Pew and Teller's. I foreshadowed the immediate
explanation earlier: Dennett denies that determinate, imagistic phenomenal experience
exists. In particular, he denies that qualia exist. Qualia include the concrete, experienced
content of a red sensation, which is and has been the very object of psychophysical laws
and linking propositions. Dennett favors instead a view according to which we attribute
qualia to ourselves as part of a narrative redescription of the information we receive
perceptually. In his view, qualia are just the subject matter of fictional stories, having the
same status as Santa Claus or the Three Bears. In technical terminology, they are
nonexistent intentional objects of narrative linguistic descriptions. But why does Dennett
deny qualia? We need a deeper explanation. This is especially demanded, since his
account of qualia as objects of narrative is an instance of a tendency he otherwise
bemoans: the tendency to linguify the psychology content. In an article in 1988 and in
his 1991 Dennett offered some arguments meant to discredit qualia, that came in the form
of thought experiments, namely about taste and vision. Now I myself don't find the
arguments compelling, for reasons of experimental design. I think they fall prey to
confounded variables including uncontrolled response bias and memory effects. But I
don't think these are his main arguments. In fact, his denial of qualia long predates these
particular arguments. It's one of the most persistent features of his writing, spanning 30
years. We could perhaps trace it to his early admiration of AI models, which tend to
linguify, or to an unavowed remainder in his admiration for behaviorism, but I think that
would be unfair and not very interesting. I think the real explanation comes in what I
would guess to be Dan's most repeated rhetorical charge against qualia, that they are
mysterious and immaterial, unreconcilable with natural science. Those who posit qualia
are in fact unable to say how those qualia could be produced by or be identical with, the
activity of neurons in the brain, that is, with the differential flow of ions across cellular
membranes in accordance with an electrical potential. But, Dennett reasons, the brain is
made of matter. The red experienced in a red sensation is not in the ontology of physics,
although the ontology of physical light is. So while the brain may carry information
about red light, it can't contain or produce red qualia; a properly naturalistic ontology
won't allow him. This line of reasoning contains two interrelated assumptions that are
interesting, widely shared, and by my lights wrong.
The first assumption concerns the domain of the natural. How shall we decide what is
natural? One was is to let the constituents of things decide. A thing is natural if it is
made of matter, and so its states and properties are in principle describable in physical
language. This contrasts with other putative objects not made of matter, such as souls or
God, which are supernatural. Let us call this "ontological naturalism." It is a version of
what Larry Shapiro has called "Lego naturalism": natural is, as natural is made out of
properly certified physical building blocks. This contrasts with what might be called
"scientific naturalism." According to scientific naturalism, a thing is natural in virtue of
being described by natural scientific law, or counted among the objects of natural
scientific explanation. The natural sciences are specified by a list, under this conception,
which always includes physics, chemistry, and biology, and often includes psychology.
On this view, it's one of the great things about Penn, in fact, that psychology is here
classed among the natural sciences. On this view, if qualia are the object of
psychophysical laws, they are natural phenomena, as long of course as one considers
perceptual psychology to be natural science.
The second assumption concerns the proper way to conduct science. Shall we see science
primarily as the search for lawful relations, and so include the laws of psychophysics in
the proper domain of science, or shall we conduct science by seeking explanations of
things in terms of what they are made of, in terms of their constituent parts? I think
Dennett favors the latter approach, given his repeated discussions of what things are
made of, and his use of what things are alleged to be made of as a criterion for
distinguishing the real and the unreal. The assumption that a real explanation involves
specifying an underlying mechanism built out of material parts is deep, but not universal.
It was encouraged by early proponents of modern science in the 17th century. It was
especially prominent in the physical explanations of Descartes and Robert Boyle, who
argued that real explanations must be cast in terms of the shapes and motions of the parts
of things, like the shapes and motions of the parts of a machine. Let us call this
mechanistic mode of system of science the "Boylean" model. Dennett's two assumptions
are deeply embedded in the history of science, or at least in a certain reading of the
history of science, that is ontological naturalism and the Boylean model. It is widely
held, that the Scientific Revolution in the 17th century had the effect of banishing the
mind from nature, and relegating mental phenomena to an immaterial mind nature in the
seventeenth century, but a larger group of investigators considered mind to be part of
nature. This included many dualists, who saw no reason why the states of an immaterial
substance could not be studied empirically, and so made an object of empirical science.
And this kind of naturalism about the mind and empiricism toward the mind, I think,
informed the work of Benjamin Pinkel, one of those for whom these lectures are named.
The new scientific study of mind was slower to gel than the new physics, and it was only
in the 18th century that substantial works in the new scientific psychology appeared.
Early works in the 1750’s were by the Swiss naturalist Charles Bonet and the German
physician Johann Kruger. As the century wore on the number of works in psychology
became more numerous. Interestingly, in 1808 when FA Karess wrote his History of
Psychology he could discuss 125 authors that he considered to belong to the community
of psychologists. Across time, the tendency within this group of literature was to defer
ontological questions about what the mind is made of, in order to study the empirical
properties of the mind, so that even the dualists deferred the metaphysical questions. The
strategy of 18th century authors to bracket ontological questions in favor of seeking
lawful relations is in the spirit of another style of science that was beginning to supplant
Boylean science at the time. The second style of science posited explanatory laws, even
where no proper Boylean or material building block ontology could be found. The most
famous was Newton's postulation of a law of universal gravitation, in the absence of any
mechanistic explanation of how the attractive force worked. Newton himself hankered
after a mechanical explanation of gravity in terms of the pushes and pulls and contacts of
particles. Others took the lesson from his writing that laws are equally or more important
that stories about constituent parts, so let's dub this second style that looks for laws rather
than constituent parts, Newtonian science.
This brief bit of history gives us a way to understand Dennett's objection to qualia, and a
way to describe Pew and Teller's work. Dennett subscribes to the tradition that sees mind
outside of nature. If mind is to be brought into nature it must be done carefully, by
equating mental activity with the activity of the material brain. We must explain mind in
terms of constituent mechanisms, for which we are able to see, in principle, how their
operation could be explained by physics. We must therefore be able to go from (unclear:
the attential?) to the design ultimately to the physical stance, in his words. If we now
can't see how to specify the component parts equated with some mental phenomena,
these phenomena should be jettisoned. This is the road of ontological naturalism in
Boylean science. By contrast, those who take psychological science such as
psychophysics seriously can describe themselves as scientific naturalist, who accept the
Newtonian model. Of course, these persons might also look for Boylean explanations
where they can be found. But they won't look for Boylean explanations exclusively, and
they'll perhaps conjecture that in many areas of science it's the Newtonian style that rules.
In the end, my main objection to Dennett's talk is his exclusionary metaphysics. He
wants to rule qualia out of present science, because it's hard to see how they can be
explained by appeal to the physical properties of the brain. In effect he is supposing that
we should extrapolate current knowledge of the brain and current physics into the future,
and predict that qualia will never be explicated, and concludes from this that they should
be denied existence. As a Boylean, he is ready to rule out what can't be explained in
Lego fashion. By contrast, the approach I advocate takes seriously the lawful in the
phenomenon. It approaches the mind on all fronts, by studying psychological capacities
and processes in their own rights, by studying their relation to physiology, and by
drawing on evolutionary findings where available. But it insists that the science of
perceptual psychology can proceed without needing a prior ontological certification,
based on current knowledge of the brain and current physics. Posited qualia are justified
by their incorporation into successful science. Psychology is an autonomous science. It
can proceed in advance of neuroscience, while listening to neuroscience bring any news.
Further reflection might even show that neuroscience needs psychology more than
psychology needs neuroscience, for the reason I gave earlier: psychology provides the
functional language for describing brain function, but that's another topic for another
philosophical age.
Dan Dennett
I'm really glad that [Hatfield] drew your attention to Teller and Pugh's concept of a bridge
locus, and he's exactly right that this is contrary to what I've claimed about qualia. If you
want to see a beautiful demolition job on the data in support of the Teller and Pugh
notion, look at Thompson and Palacio and Varela's target article in Behavior and Brain
Sciences of 1992, in which I also have a commentary.7 I think that we can't do justice to
this issue here, but I think it is really interesting to see why you don't need a bridge locus
to do psychophysics; something that Thompson, Palacio and Varela explain in some
detail. As for the distinction between Newtonian and Boylean science, I think Hatfield's
right about the history, and I'm also going to accept his claim that I am playing the
Boylean role, but I want to point out that in doing this I am simply supposing that, just as
one can be a good Boylean about reproduction, metabolism, and locomotion (for
instance), one can be a good Boylean about psychology. I don't see any reason to
suppose that the phenomena of psychology will require a different attitude towards the
natural sciences than the other, initially deeply perplexing phenomena of life. If we go
back in time, not so very far, we find that the mystery of reproduction was one that was
so great that preformationism reigned, or was at least the most serious contender for a
theory, and people hadn't a clue how using mechanisms you could explain the process of
reproduction. We now have a very good detailed Boylean mechanistic explanation of
how reproduction is possible. Our aspirations in cognitive science should be the
aspirations to treat the mind the way we treat the rest of the body, as a very complex bit
of machinery that will succumb to reverse engineering. So I accept Hatfield’s distinction
between Newtonian and Boylean science and I say, as far as psychology is concerned,
let's be good Boyleans.
Hatfield
Well, the only response I would make is that for a convincing defense of the cogency of
the Pew-Teller approach, see the forthcoming, still in preparation article by Hatfield and
Pew appearing soon near you. Other than that, I would just argue that the Newtonian
approach should be pursued as well.
Seyfarth
Well, all this is going to do is serve to show the different ways in which philosophers and
psychologists either prepare themselves or don't prepare themselves for being a
commentator. I should start off by saying that I work on the social behavior and
communication of animals, and I think Dan's paper makes a very cogent argument for the
integration of lots more work on nonhuman creatures into the central part of cognitive
science. Of course I believed this before I heard his paper, so in that sense he had no
effect on me whatsoever.
When one embarks on a program in artificial intelligence, as Dan was describing, one
sets up a structure in the program that embodies in most cases the two kinds of
assumptions that he's criticizing, and hence as he argues, quite well I think, you recreate
the problem because you're working within a specific framework. Now consider the
difference between this kind of research and the sort of research one does when one goes
out to study an animal in its natural habitat or in the laboratory. There, you take as given,
take for granted, that you're dealing with an evolved structure. There are two things
about this evolved structure that are important. First of all, you know that it's going to
have a lot of gerrymandering and ad-hockery, because we know that's how evolution has
worked over the animal's history. Second, you can't isolate the brain from the rest of the
animal. It forces you to integrate all of the biological systems in order to explain how the
animal achieves what it does. This is certainly what has happened in the course of our
understanding of the mechanisms that govern reproduction in vertebrates or invertebrates.
But this also forces us to entertain the possibility that there are many different answers to
questions about the proper functioning of mind. Dan gives us three questions that he
says, and I think Gary is right, these aren't, maybe not be the right questions, it might be
better to say they're not the only questions, and he says what we've got to do, in the future
of cognitive science, is to ask how else we might conceptualize the proper parts of a
person.
I would suggest that a lot of the reason why we frame the questions in this way is because
so much of cognitive science deals exclusively with humans. Imagine that I had sitting
here a research subject, a salamander. Would I be prompted to ask, do facts about the
background have to pass through belief fixation in order to influence planning or is there
a more direct route from world knowledge? Probably not. I'd be forced to ask some
other kind of question that nevertheless gets at the same general biological problem. So
what I take this as, is an argument very strongly in favor of taking taking a large part of
the Institute's space and devoting to work on things like salamanders.
Having said that, I just want to ask, I just want to throw out one other idea. Every year,
the Boston Museum of Science, I think it's every year, has a contest in which people are
asked to submit programs that judges, the programs talk with you on a computer terminal
and the judges sit at the end, the other end of the computer terminal, and they're supposed
to figure out whether it's a computer program or a real person at the other end, and I
know that Dan has been the judge of this contest one year, maybe it's not every year
maybe it's every so often. I am also pretty sure that members of the general public are
also invited to be judges. Here's an example in which a person is presented with a mind,
or a candidate mind, and chances are, if it's a member of the general public there would
be an attempt to linguify mental content, to structure it in the way that we in cognitive
science tend to do, and similarly in the Cog project, what matters to the success of that
project in some respects is whether Kismet elicits this parental response. Not how it does
it. So projects like these, like ethology in animal behavior, force you to use function as a
guide rather than structure. We want to know how an animal achieves its success in the
world before we start talking about the structure, and I think this is an important sort of
shift of emphasis and it's certainly an emphasis that involves more biological
approaches. A lot of this is what Dan Dennett has been saying for years. He wrote a
commentary in Behavioral Brain Sciences many many years ago, in which he chastised
people in artificial intelligence for using fancy computers to try and model one particular
tiny part of the human brain. He said, there's a much, much more difficult problem, and
he titled his article, "Why not the whole iguana?" And I think that is a much more
difficult problem, and it's something that we ought to start tackling.
Dennett
There’s very little for me to disagree with there. Indeed, "why not the whole iguana?"
This has turned out to be a very fruitful research strategy within artificial intelligence and
artificial life.8
A little story about the early days of the Turing Test may shed some light on some of
Robert’s other comments. The recent limited Turing Test competition, for the Loebner
Prize, was held not at the Museum of Science but (the first year) at the Computer
Museum in Boston, and I was chairman of the Prize Committee over for several years.
Some years before that, Joseph Weizenbaum, another member of the Prize Committee
and the creator of the famous or notorious Eliza program (the Rogerian psychotherapist
that interviews you about your psychological problems) ran a little informal experiment
at Harvard Medical School. He wanted to see what would happen if they confronted
people with something like a Turing Test, so he introduced subjects (psychiatry residents
as I recall, but I’m not certain) one at a time to a human being and said, “you're either
going to be talking to this human being or you're going to be talking to a computer.”
And the human being would shake hands with the subject, and they'd make a few
pleasantries perhaps, and then the human being would go off, and then the subject would
sit down at a terminal and start a question-and-answer game with . . . either the human
being in another room, or Eliza, the computer program. After a few minutes of
interaction, Weizenbaum asked the subject in passing what his opinion was so far: did he
think he was conversing with a person or a computer. If the subject gave the wrong
answer, Weizenbaum would surreptitiously adjust the performance: if the subject said he
was conversing with a computer, the actual human interlocutor would raise the quality of
his responses, making them ever “more human” and if the subject said he was conversing
with a human being, Weizenbaum would degrade Eliza’s performance. Subjects were
remarkably perseverative in their hypotheses, clinging to them in the face of ever
mounting evidence to the contrary. In one instance the human interlocutor was driven to
the point of saying something like, "I really liked that blue necktie that you had on when I
shook hands with you a few minutes ago." In one case in which a subject thought Eliza
was a human being, Weizenbaum found that when he degraded the performance of Eliza
to the point of "word salad", the response from the subject was, "I don't know, I think this
man's sick, I don't know what's on his mind." The fact is that we are much more
susceptible than we are prepared to acknowledge to the intentional stance. Once you get
into that frame of mind, it’s child's play to make the world fit the hypothesis; it's child's
play to find reasons, to find intention, in almost any behavior that emerges. A hard thing
to do in cognitive science is both to exploit the intentional stance and also resist its allure
when it should be resisted.
Lecture Notes
1. A shorter version of this talk has been published in The Foundations of Cognitive
Science, Joao Branquinho, ed., Oxford University Press, 1999. I want to thank Chris
Westbury and Rick Griffin for comments on an earlier draft.
2. The previous 6 paragraphs are drawn, with some revisions and additions, from my
Kinds of Minds (1996).
3. Cynthia Brezeal, discussion at American Association for Artificial Intelligence
Symposium on Embodied Cognition and Action, MIT Nov, 1996.
4. Consider a sort of problem that often arises for learning or problem-solving programs
whose task can be characterized as “hill-climbing”--finding the global summit in a
problem landscape pocked with lower, local maxima. Such systems have characteristic
weaknesses in certain terrains, such as those with a high steep, knife-edge “ridge” whose
summit very gently slopes, say, east to the global summit. Whether to go east or west on
the ridge is something that is “visible” to the myopic hill-climbing program only when it
is perched right on the knife-edge; at every other location on the slopes, its direction of
maximum slope (up the “fall line” as a skier would say), is roughly perpendicular to the
desired direction, so such a system tends to go into an interminable round of
overshooting, back and forth over the knife-edge, oblivious to the futility of its search.
Trapped in such an environment, an otherwise powerful system becomes a liability. What
one wants in such a situation, as Geoffrey Hinton has put it, is for the system to be
capable of “noticing” that it has entered into such a repetitive loop, and resetting itself on
a different course. Instead of building an eye to oversee this job, however, one can just let
boredom ensue.
5. Antonio Damasio's recent book Descartes' Error (New York: Grosset & Dunlap, 1994)
is a particularly effective expression of the new-found appreciation of the role of
emotions in the control of successful cognition. To be fair to poor old Descartes,
however, we should note that even he saw--at least dimly--the importance of this union of
body and mind:
By means of these feelings of pain, hunger, thirst, and so on, nature also teaches that I am
present to my body not merely in the way a seaman is present to his ship, but that I am
tightly joined and, so to speak, mingled together with it, so much so that I make up one
single thing with it. (Meditation Six)
6. In what follows I owe many insights to Lynn Stein's concept of "post-modular
cognitive robotics" and Eric Dedieu, "Contingency as a Motor for Robot Development"
AAAI Symposium on Embodied Cognition and Action, MIT Nov, 1996.
7. "Hitting the Nail on the Head," (commentary on Thompson, Palacios and Varela),
Behavioral and Brain Sciences, 15, 1, p. 35, 1992.
8. In July of 2000, I participated in an international workshop on the island of Lanzerote,
in the Canary Islands, entitled “Towards the Whole Iguana.” A volume of the papers and
discussions presented by the roboticists and artificial life researchers who participated,
edited by Owen Holland and David McFarland, is forthcoming from Oxford University
Press.