What kind of machine could truthfully affirm the Cartesian cogito for itself?

The famous statement cogito, ergo sum, "I think, therefore I am," expresses a profound ontological insight.

Although it may be impossible to know for certain that the world is other than an elaborate hallucination, I know at the core of my subjectivity that I do experience this world, whether a dream or not, and therefore the "I", or homunculus that experiences it must exist.

Something must exist to have the thoughts and experiences that I have, but what precisely does it mean to "have experiences"?

Descartes reasoned that a self must exist if it knows that it has experiences, even when what is experienced cannot be proven to exist. The riddle becomes knottier in the case of a machine that could sense the fact of its own sentience.

A machine can certainly be sensitive and responsive to its environment and to many of the conditions internal to itself, but we know that this is just programmed electro-logical cause and effect.

One might also argue that a brain is also just responding organo-mechanically to its environment, but then where is the "I" that I personally and subjectively knows that it exists?

One may convince oneself that other persons are merely complex response mechanisms, but one may never escape the innermost Cartesian conviction that the "I" to which experiences are happening is much more than a flurry of brain processes.

Even conceding that the brain has a mechanism for prioritizing all sense data into degrees of selflikeness, the Cartesian conviction resists admitting that the "I" is actually the brain's categorization mechanism.

Do we mean that the brain processes are saying "I think, therefore I am"? No, because that would personify what we have already claimed was a mechanism. If the mechanism is personified, then it would also need its own private cinema of experience, and where would the categorization mechanisms within the categorization mechanism derive their Cartesian conviction?

Mental states, such as those attained through religious meditation or by taking psychelics, in which the sense of the self as a distinct recipient of experience relaxes or apparently even subsides entirely, still uphold the Cartesian conviction.

Commonly heard descriptions of these states are "feeling totally one with the universe" or "having a God's-eye view". In such cases, it is evidently "God" or "the Universe" who affirms that "I think, therefore I am"?

A machine, on the other hand, does not need to epiphanize that it is at one with the universe.

Consider a machine that was not only sensitive and responsive to its internal and external environments, but that was also able to respond to the state of the mechanisms that sensed its internal and external environments.

This machine would not only process data about the state of its environment and about the state of certain of its internal operations, it would also have a special mechanism monitoring its processing of those states.

The sense-mechanism-monitor can process data about the state of the data processing mechanisms, but no mechanism would monitor the state of the sense-mechanism-monitor. (Zen teaches that that no one can see the original face.)

The sentient mechanisms watching the environment might be equivalent to a "self". If the sense-mechanism-monitor had a feedback circuit with the sentient mechanisms, then the sentience could query the sense-mechanism-monitor.

The message received from the sense-mechanism-monitor would interact with the sentience, or "self", like input from any other internal or external signal. That's the key. The self receives signals from its own monitor as if it were external data.

Our homework assignment is to try to convince ourselves that "I think, therefore I a neural feedback circuit am."

When we finish that assignment then we'll have to find out who convinced whom.

David Chalmer's Home Page

Michael Webb, 2002

home ]