Gabor

Accent Adaptation

Refixation

Gabor

As an individual reads, visual information is being collected each word's identity from the letters on the page. Current models of (visual) letter recognition account for either letter order effects or perceptual similarity effects. Here, we aim to model human visual letter recognition by incorporating known neuro-biological shape/form response profiles of the primary visual cortex. Gabor-based wavelet convolution, a computer vision technique, successfully performs object recognition and is a candidate for capturing human-like letter order and perceptual similarity effects. This technique extracts higher-level features of an image using kernels with response patterns mimicking simple cells of the primary visual cortex.

learn more
associated git repo
Accent Adaptation

With adequate experience, listeners improve their ability to comprehend accented speech. Previous work demonstrates that listeners who adapt to one accented talker generalize that adaptation to other accented talkers - exposure to multiple talkers of the same accent facilitates comprehension of a novel talker of that accent (Bradlow & Bent, 2008) and exposure to multiple novel accents facilitates comprehension of yet another novel accent(Baese-Berk, Bradlow, & Wright, 2013). This project introduces a model of accent adaptation in which the task of accent adaptation is represented as a problem of hierarchical Bayesian inference, which assumes that listeners simultaneously learn about the distribution of talkers, accent groups, and accented speech more generally.

learn more
Refixation

By studying where refixations go, this study aims to address how visual information is used in word identification.

learn more