What is it about?

We study how people's eyes move over a text in order to figure out how their minds create meaning from language. We find that the time spent reading a word is a logarithmic function of how predictable the word is in context, which means that small differences in low predictability (e.g., between 0.01 and 0.001) create large differences in processing difficulty. This is hard to explain under the widespread view that prediction makes comprehension easier by preactivating words early, but it makes sense if prediction instead reflects a probabilistic process of managing uncertainty about the meaning.

Featured Image

Why is it important?

There has been a long and confusing debate about what causes predictability effects during reading and what these effects imply about our comprehension processes. Our study pushes toward settling a key aspect of this debate (the functional form of predictability effects) by revisiting it with more data, more powerful models, and more stringent tests than any previous work.

Read the Original

This page is a summary of: Large-scale evidence for logarithmic effects of word predictability on reading time, Proceedings of the National Academy of Sciences, February 2024, Proceedings of the National Academy of Sciences,
DOI: 10.1073/pnas.2307876121.
You can read the full text:



The following have contributed to this page