Sam Norman-Haignere (@SamNormanH) profile picture

I study the neural computations of human hearing with a focus on speech and music perception. @statsROC @URNeuroscience

Joined April 2017
Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)Photo by Sam Norman-Haignere (@SamNormanH)
New preprint from David Skrill with @jenellefeather: tinyurl.com/2xm8rucz We synthesize stimuli that decorrelate neural predictions from competing encoding models, revealing a dramatic improvement in auditory cortical prediction accuracy from adversarial training.
1
5
13
765
The robust and standard models made highly divergent predictions within this shared neural subspace, providing an interpretation for how prediction decorrelation was achieved.
1
17
We think the overall approach described here is simple, effective, and general. We hope others will find the ideas presented here useful in their own research!
9
Human auditory cortex integrates information in speech across absolute time (e.g., 200 ms), not phonemes, syllables, words, or any other time-varying speech structure: nature.com/articles/s41593-0…
4
18
2
58
7,386
We replicate this result using naturally fast and slow speech.
1
1
158
How does the human brain recognize a word using time-yoked auditory computations? We’re working on that now.
4
145