Connections, Symbols, and the Meaning of Intelligence

More recently, debates in AI have focused on the implications of Connectionism. Connectionism is the hypothesis that distributed computations are capable of instantiating intelligent functions without relying on the representational character of symbols, but rather on the computational states themselves which are cal1~ distributed representations (Haugeland, 1991). This distinction puts connectionism at odds with symbolic theory. The current debates tend to be over which theory will yield intelligent systems--symbolic or connectionist? But as we will soon see, this really amounts to a debate over which representational scheme is required for general intelligence