Spatial knowledge during skilled action sequencing: Hierarchical versus nonhierarchical representations

Typists can type 4 to 5 keystrokes per second at around 95% accuracy, yet they appear to have poor declarative knowledge of key locations. Logan and Crump (2011, Psychology of Learning and Motivation, Vol. 54, pp. 1–27) accounted for this paradox by proposing that typing is hierarchically organized into two loops, with an outer loop that transforms sentences into words and passes each word, one at a time, to an inner loop that transforms each word into its constituent keystrokes; however, the nature of the inner loop’s spatial knowledge is not well understood. Key locations may be learned through the experiences of locating and traversing between keys. In daily life, people tend to type structured language, and, as a consequence, certain keys and key-to-key transitions are experienced more frequently than others. Here, we asked whether or not this knowledge is structured hierarchically. For example, knowledge of key locations may be nested within representations of words, or the inner loop may rely on knowledge that is independent from higher level structures. To test this, we had people type English, English-like, and random strings during normal, partially occluded, and occluded typing. In both partially occluded and occluded typing, error rates were higher while typing random strings compared to English and English-like strings, whereas there was no difference in error rates between English and English-like strings. This suggests that typists’ spatial knowledge of the keyboard is not driven by hierarchical word-level representations, but instead is likely driven by a collection of individual processes, such as knowledge of the sequential structure of language acquired by typing more frequently occurring letters.