Back

A compositional shape code explains how we read jumbled words

Agrawal, A.; Hari, K. V. S.; Arun, S. P.

2020-04-16 neuroscience
10.1101/653048 bioRxiv
Show abstract

We read words and even jubmled wrods effortlessly, but the neural representations underlying this remarkable ability remain unknown. We hypothesized that word processing is driven by a visual representation that is compositional i.e. with string responses systematically related to letters. To test this hypothesis, we devised a model in which neurons tuned to letter shape respond to longer strings by linearly summing letter responses. This letter model explained human performance in both visual search as well as word reading tasks. Brain imaging revealed that viewing a string activates this compositional letter code in the lateral occipital (LO) region, and that subsequent comparisons to known words are computed by the visual word form area (VWFA). Thus, seeing a word activates a compositional letter code that enables efficient reading.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.