In this fascinating piece, the philosopher Stephen Asma lays out a compelling case for why the faculty of imagination may be older - and conceptually prior - to language. In his formulation, wordless imagination was an earlier step above the "lizard brain" state of purely stimulus-response and instinctual behavior. It represented the ability of the brain to generate states that were not triggered directly by stimuli, i.e., a kind of confabulation. Language is then seen as a further complexification, where an apparatus evolved to harness the power of imagination to more useful ends.
This view resonates strongly with my own conception of the nature of the mind, which must be seen in an evolutionary and developmental context to be fully understood. The the "higher" capabilities of human and animal cognition are essentially layers built on top of the basic stimulus-response animal. As these layers of intertwining and autonomously activatable neural networks developed, they created increasingly complex ways for information to "dwell" and ramify within the brain, thus creating room first for imagination and then for language. And next, who knows for what?
This is especially significant in the quest for AI. The entire AI enterprise began with a top-down symbol-centered view. Even though that has now given way to more emergence-based conceptions such as neural networks and subsumption, the phylogenetic sequencing by which intelligence arose in the world is still not appreciated. Nor is the similar ontogenetic sequencing whereby a developing animal moves from a "primitive" stimulus-response functionality towards an increasingly "linguistic" one based on explicit communication. This is true not just in humans but also in primates, birds, probably cetaceans and many other vertebrates. Perhaps we need to abandon all the "short cut" methods we are currently using to produce artificial intelligence, and follow Nature's evolutionary pathway. Start with the simplest Braitenberg vehicles, and let them evolve brains capable first or imagination and then of language. It would be especially interesting to see whether these things emerge spontaneously as a consequence of complexification in brain and body.
This view resonates strongly with my own conception of the nature of the mind, which must be seen in an evolutionary and developmental context to be fully understood. The the "higher" capabilities of human and animal cognition are essentially layers built on top of the basic stimulus-response animal. As these layers of intertwining and autonomously activatable neural networks developed, they created increasingly complex ways for information to "dwell" and ramify within the brain, thus creating room first for imagination and then for language. And next, who knows for what?
This is especially significant in the quest for AI. The entire AI enterprise began with a top-down symbol-centered view. Even though that has now given way to more emergence-based conceptions such as neural networks and subsumption, the phylogenetic sequencing by which intelligence arose in the world is still not appreciated. Nor is the similar ontogenetic sequencing whereby a developing animal moves from a "primitive" stimulus-response functionality towards an increasingly "linguistic" one based on explicit communication. This is true not just in humans but also in primates, birds, probably cetaceans and many other vertebrates. Perhaps we need to abandon all the "short cut" methods we are currently using to produce artificial intelligence, and follow Nature's evolutionary pathway. Start with the simplest Braitenberg vehicles, and let them evolve brains capable first or imagination and then of language. It would be especially interesting to see whether these things emerge spontaneously as a consequence of complexification in brain and body.
No comments:
Post a Comment