Carnegie Mellon | School of Computer Science





Manuel Blum

As a kid, I was the ANTITHESIS of a PRODIGY. I was a dumb kid that wanted to be smarter. When I was 6 years old it occurred to me that if only I understood how brains work, I could perhaps be a bit more intelligent.

In college, I studied Electrical Engineering in the hope that from resistors, capacitors, vacuum tubes and motors, it might be possible to build a mechanical insect.

As a junior, I apprenticed myself to the emminent neurophysiologist Dr. WARREN S. McCULLOCH.

WARREN WAS GREAT. He was warm. He was loving. He knew as much as anyone did about the brain. His paper with Walter Pitts had introduced a simple model of a neuron, then proved that automata constructed from these neurons and supplied with a paper tape could simulate a Universal Turing Machine. Neurophysiologists were skeptical. The McCulloch-Pitts neuron required both excitation and inhibition, and the neurophysiologists of the time had not detected inhibition. McCulloch and Pitts insisted it must be there, as they had shown it necessary to construct a Universal Turing Machine.*
* A few years after McCulloch and Pitts predicted it, the neurophysiologists did indeed find inhibition in the brain.

I learned a lot about the brain in Warren's lab. From his student, Jerry Lettvin, I learned that the eye turns off during a saccade. Helmholtz had pointed out we couldn't read otherwise. From Edwin Land, I learned that the eye can see a rainbow of colors when shown just two different frequencies of red.

In 1969, I became an Associate Professor of Electrical Engineering and Computer Science. Tenure gave me the freedom to do something strange. With Lenore's blessing, I took a week off to go to a BLISSFULLY QUIET Berkeley Hotel to try to understand/retrospect on WHAT IS CONSCIOUSNESS.

And what did I learn? In a word, I determined that consciousness is like a FLASHLIGHT that one can direct and focus on whatever one finds interesting. And when one does so, IDEAS BUBBLE UP. From where I do not know, but bubble up they do.*
* These insights take me surprisingly long to unearth, and remind me of Kurt Vonnegut's "Breakfast of Champions." There, a brave and selfless soul makes enormous sacrifices to learn... mere trivialities.

I have since learned something else that is quite surprising about consciousness: that what we are conscious of is not something that we can close our eyes and draw or describe in any detail. Rather, we can describe what we are conscious of only roughly. On the positive side, we can focus on any facet of it that we so choose -- any facet that we desire to know more about. We have the potential to focus on and be conscious (recursively) of any part that we are asked to look at or into.


To get us thinking on the same plane, I'll state a pretty good DEFINITION from Collins English Dictionary:

4.a. denoting or relating to a part of the human mind that is aware of a person's self, environment, and mental activity and that to a certain extent determines his choices of action.
b. (as n, ): the conscious is only a small part of the mind. Compare <unconscious>

Good as it is, this definition misses a lot. It makes no allowance for the possibility that nonhuman minds may be conscious. It gives no tools for deciding if a robot is conscious. It gives no sense of how or why consciousness arises.

My approach to this problem, besides introspection, besides reading all I can about the neurophysiology and psychology of the brain and the philosophy of consciousness, is to think through the computer science of the problem -- to try and clarify what I don't understand through the making of models, definitions, through reasoning and by coming to grips with the paradoxes.

There are two different views one can take of a HUMAN BEING:
1. the view that the human is a MECHANISM, an automaton with substantial but finite internal memory, programmed like any computer to do whatever it does, and/or
2. that the human is a thoughtful observant creature with a God-like free will; that it is a CONSCIOUS ENTITY at the controls of a highly complex highly capable mechanism, choosing what to do from among options served up by/from its vast unconscious below.


As Samuel Johnson (1709-1784) pointed out:
All theory is against the freedom of the will;
all experience is for it.

From the mechanistic point of view, there is nothing immoral about causing pain... yet it is clearly immoral to cause a conscious entity to suffer. So the mechanistic view is missing something!

ANOTHER PARADOX: Only a tiny very high-level fraction of our brain is conscious. What makes that part and only that part conscious?

For example:

1. We don't know from where ideas bubble up -- though we happily take credit for those that do.

2. We have little say about what data structures the brain uses. These decisions are made unconsiously.
(For example, you have memorized the alphabet. How is that info stored in memory? Is it singly-linked, doubly-linked, or random access? Actually, none of the above. And how is language stored? Differently in simultaneous translators than in people who speak two languages.)

3. Breathing is interesting because it can be either conscious or unconscious.

4. Look at the world, close your eyes, and try to visualize that world. Your mind's eye has a hard time doing it.

5. You can see the world quite well in your dreams. How come your conscious mind can't do this for itself?

6. Try to describe your mother's face to someone. You'll have a hard time doing this. It's not just a matter of language. You can't draw her either. Yet you have no trouble recognizing your mother when you see her! You can pick her out from among thousands of people passing through immigration. You can do so even though you haven't the words, the drawing ability, or the memory to describe her so that someone else who has never seen her can pick her out.

Our brain does much unconsciously. What is the difference between what it does of which we are conscious and that of which we are unconscious?

7. We are UNconscious of how we ride a bicycle. Do you know how we make a left turn on a bike?

8. We are UNconscious how our eye turns off and on during a run, in time with our footsteps.

9. When you put off doing something important, it starts to gnaw at you. What is the "it" that gnaws at you? What is the "you" that is being gnawed at?
That "something important" wells up to plague you for as long as you put it off. It can give you "ulcers" if you don't attend to it.

I can imagine a mechanism -- a robot -- having self-awareness, that is, a representation of itself in its environment, as that self-awareness is essential for prediction. What I'm missing here is the step from self-awareness to consciousness, but the two concepts are close. In my theory, self-awareness is a big step toward consciousness.

How does one tell if an animal is or is not a conscious entity? How does one tell if a dog, a rat, a frog or an ant is conscious? How can one argue -- without referring back to oneself -- that humans are conscious entities? My back reference argument for your consciousness goes like this: I personally have experienced consciousness: I know that you are constructed like me. So I conclude that you, like me, experience consciousness.

What is needed is:
a definition of consciousness that can be used to decide which animals are conscious and to what extent.
a definition that goes beyond "I am conscious; you are built like me; therefore you are conscious."
a definition that will enable us to determine whether and to what extent neurons (more generally living cells) are conscious.
a definition that will enable us to determine whether and to what extent the community of all human beings, as a society, is a conscious entity.
a definition that will enable us to decide: are ants conscious? ant colonies?
a definition that will enable us to determine whether a robot is conscious.

I would prefer to see an argument for consciousness -- at least for consciousness in Darwinian evolved organisms -- that goes somewhat like the following argument for deciding which organisms do or do not feel PAIN:

The argument goes like this: Pain is useful.

1. Break a leg, and pain will cause one to favor that leg, thus preventing further damage. If an entity uncoerced continues to use its damaged leg (to the extent it can physically do so) without regard to the damage, then I would suggest that it feels no pain in that leg.

2. Break a leg, and pain usefully causes one to avoid a repetition of whatever experience led to the breakage. If the entity repeats the experience without learning from it, then I suggest that it feels correspondingly little pain in that leg.. because pain does it little good.

If an electrical shock has no effect on the actions of an entity, then I venture that the shock does not cause the entity any pain.
If a moth keeps crashing into a hot light bulb without regard to the damage that inflicts, then I would argue that it does not feel the heat as pain.
Similarly, a tree feels no pain because it cannot make use of the pain in any way.

Consider defining consciousness by saying that an entity has consciousness if it is aware of its environment, of itself in that environment, and if it has the potential to extend that awareness to any "desired" facet of its environment, in particular to any facet that one is asked to attend to.


Consider defining consciousness as follows:
An entity is CONSCIOUS relative to a given environment if
1. it has a model of that environment,
2. it has a model of itself as a free agent capable of choosing from among a palette of choices in that environment,
3. it can make predictions based on the actions it (freely) chooses to make for itself in that environment, and
4. it can learn from its mistakes -- to improve its predictions.


* Rodolfo Llinas "I of the vortex: From Neurons to Self"
MIT Press. 2001

* Antonio Damasio "The FEELING of WHAT HAPPENS: Body and Emotion in the Making of Consciousness" Harcourt Brace & Co, 1999

* Daniel Wegner "The Illusion of Conscious Will" MIT Press, 2002

* Scientific American Special Edition: The Hidden Mind,
On the newstands until 31 August 2002. This issue has many wonderful Includes:
Antonio Damasio "How the brain creates the mind"
David Chalmers "The Puzzle of Conscious Experience