Connectionism and the problem of systematicity (continued): why Smolensky's solution still doesn't work.

Paul Smolensky has recently announced that the problem of explaining the compositionality of concepts within a connectionist framework is solved in principle. Mental representations are vectors over the activity states of connectionist "units", but the vectors encode Classical trees, whose structural properties in turn "acausally" explain the facts of compositionality. This sounds suspiciously like the offer of a free lunch, and it turns out, upon examination, that there is nothing to it.