systematicity

The property whereby a mind's capacity to entertain certain structured thoughts, or to understand certain sentences, comes bundled with the capacity to handle others built from the same parts: anyone who can think that John loves Mary can think that Mary loves John. Jerry Fodor and Zenon Pylyshyn argued that classical symbol systems, being compositional, explain systematicity automatically, whereas connectionistconnectionism, n.: The view that minds and intelligent machines are best built from huge networks of simple, neuron-like units whose connection strengths are learned, rather than from explicit rules and symbols. networks either fail to exhibit it or succeed only by implementing a classical architecture. In machine learning the challenge became the study of systematic or compositional generalization — whether models trained on some combinations of known elements handle new combinations — tested by benchmarks such as SCAN (2018), on which standard sequence-to-sequence networks failed. Brenden Lake and Marco Baroni showed in 2023 that a network trained by meta-learningmeta-learning, n.: Training a system across many different tasks so that it gets better at learning itself, letting it pick up a new task quickly from only a few examples. for compositionality could match human systematic generalization; how systematic large language modelslarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code. are remains debated.