This proposal requests funds to partially support a symposium on Architectures for Intelligence. The major purposes of the symposium are: (1) To promote interaction among researchers who are pursuing the architectural question from divergent viewpoints. (2) To exhibit the common issues in architecture research that may have been obscured by the variety of approaches. (3) To see if there are a common set of good ideas that crop up in a variety of architectures. (4) To compare varying degrees of ontological commitment, which range from an architecture is just a notation for computations, and any convenient one would do as well to there is one optimal architecture, both for the human mind and the artificial mind, and our architectures are hypotheses about what that real architecture is. (5) To examine the levels of description idea, which is used, for instance, to say that connectionist architectures describe the mind at a finer-grained level of description than serial, symbolic architectures, so both descriptions can be right at the same time. The speakers at the symposium have all written chapters for a book entitled Architectures for Intelligence, which is being published by Erlbaum and should appear in March, 1991. (sdw)