While lying in bed this Sunday morning a few thoughts on consciousness came to me. Morning insights can be useful or vapid — not sure which these are. But they’ve stuck in my head, like a tune that keeps replaying. I’d like to share them and discuss them. Three semi-awake assertions:
- A conscious agent must be able to make a statement of fact
- Consciousness is an act of communication
- The statement of fact cannot be the state itself; it must be a symbolic representation of state
Added about 10 minutes later:
- Meaning is the ability to translate the code to a representation of the state. That is, the message must be able to be encoded and decoded — it must include a transform and a reverse transform. Such a reverse decoder must be able to make internal representations of external state “feel” like the external state. Hence, the necessity of qualia.
- To demonstrate that these internal representations are true representations of external states there must be conditions in which they can become synchronized with, and predictive of, external states; for example, voluntary action.
item 1, ‘a conscious agent must be able to make a statement of fact‘. Let’s start with a photodiode and see why it isn’t conscious. One might imagine that the diode could make one of two statements: “I am on” or “I am off”. This is impossible for two reasons. First, for the diode there is no “I”. “I” requires a separation of self from non-self, and there’s no reason to guess that the diode would know a boundary. Second, the diode has no memory. If we somehow grant that the diode can look within itself and see that it is “on”, how would it know that the current state can change, and that “off” is a possibility. It needs awareness — representation or memory — of alternate states. While an external observer might see that if electric input would turn off the diode state would change, I can’t see how the diode itself could encode alternate states. Carrying this example further, being able to make a statement of fact means that the conscious agent must be able to symbolically represent things (objects) and states of things.
Items 2 and 3. Conscious items must be symbolically encoded and thus be acts of communication. A rock falling is not, in itself, a conscious act. A statement that a rock is falling may be part of consciousness. Saying that consciousness is an act of communication carries a lot of baggage. There must be a sender, a representation and a receiver. Is this too heavy a burden? We now know that the CNS of advanced animals is more than a reflexive device. Specifically, it relies on internal representations. A reflexive device need not symbolically encode, while a representation must. We know, for example, that rat hippocampal neurons replay previous courses of action (paths taken) while the rat pauses or sleeps, and anticipates possible courses of action at choice points (paths, again). These patterns of activity fit nicely with symbolic representations, but are incompatible with a reflex action system.
If a conscious representation is an act of communication, who or where is the reader?
Items 4 asserts that, in addition to a symbolic encoder, there must be a decoder mechanism. Close your eyes and imagine your bedroom: you’ve performed an act of consciousness. The representation is in your brain. It is not your bedroom, but an activity pattern of neurons. The transform from bedroom to brain is encoding. What you “see” in your minds eye is your bedroom, or something like it. This means that activity pattern is decoded.
Item 5 tries to deal with meaning. In Searle’s famous Chinese room argument, a translator could encode and decode between Chinese and English without a sense of meaning; without knowing what the languages (codes) represent. If the internal state can not only respond to external states, but change external states and predict them, it can generate a model of external states. Generating a model of external states is how an internal representation can generate meaning. In brief, the internal representation must not only respond to external states, it must be a source of action upon external states. My guess is that there must also be a priori stuff in the internal states (Kant); things like space and time, qualia and causality.
Postscript: this is clearly very primitive, and I’ve made no attempt to connect these thoughts to the ideas of others. (I am aware of lots of other stuff). The post is an invitation for discussion.