Skip to main content
. Author manuscript; available in PMC: 2011 Jan 1.
Published in final edited form as: Cogn Sci. 2010;34(1):10–50. doi: 10.1111/j.1551-6709.2009.01076.x

Figure 3.

Figure 3

The finite-state grammar underlying the sequences used in Experiment 1 (left) and 2 (right). The sequence used in Experiment 1 has a large branching factor and a large number of highly similar paths while the grammar underlying Experiment 2 is much simpler. In contrast, the sequence tested in Experiment 1 affords a simple, first-order statistical description, while the sequence tested in Experiment 2 relies entirely on higher-order conditional probabilities. See the introduction for each experiment for a detailed explanation.