Anna Bruns¹ (), Najib J. Majaj¹, Maria Pombo¹, Minjung Kim², Denis G. Pelli¹; ¹New York University, ²Independent research scientist (Work completed while at Meta Platforms Inc.)

People read at widely different speeds, but we don’t know why. Type designers emphasize reading comfort and create “Text” fonts that are comfortable to read, but we still lack a scientific account of how letter shape affects reading. One known constraint on reading speed is crowding – the failure to recognize a letter because its neighbors interfere. Here, we report the first joint measurements of reading speed, reading comfort, and crowding. Participants completed two reading tasks: ordinary reading of a multi-page story and Rapid Serial Visual Presentation (RSVP) reading of nouns flashed serially in one location. After each story, they rated reading comfort (1 to 7). Each of 24 fonts was tested with 26 to 49 participants. Crowding distance – the minimum spacing needed to identify a letter – varied across both fonts and individuals, and explained more than half the variance in reading speed and comfort (R² = 54% for RSVP reading, 55% for ordinary reading, and 73% for comfort). Higher-crowding fonts were read more slowly, with less comfort. Higher-crowding participants also read more slowly but with the same comfort. Crowding varied sevenfold across participants, and fourteenfold across fonts. The product of font and participant crowding factors predicts 27% of the variance in individual reading speeds. These results suggest that reading is constrained by a crowding bottleneck, and that reading comfort is a perceptual marker of low crowding. The letters in our high-crowding fonts are connected, broken, or compressed, making them harder to individuate. Attempting a neuroscientific account of these behavioral results, V4 cortical crowding distance scales with perceptual crowding distance, suggesting that more V4 neurons are required to individuate letters in high-crowding fonts.

Acknowledgements: Supported by Meta, Inc., contract to D. Pelli at NYU.