speech · prosody · lm
pilm
prosody-internalized language modeling. whether a small lm can internalize prosodic structure from text alone, and what that buys for downstream comprehension.
science at the intersection of
I work between cognitive science and applied ai.
trained as a philosopher, then a neuroscientist, now a computational linguist. the questions I started with about minds turned into questions about model architecture and training data.
ms computational linguistics
ms neuroscience & psychology
fmri research fellow
ba philosophy
speech · prosody · lm
prosody-internalized language modeling. whether a small lm can internalize prosodic structure from text alone, and what that buys for downstream comprehension.
memory · retrieval
why retrieval-augmented memory forgets so badly. a two-system architecture borrowing from how cortex and cerebellum coordinate memory in mammals.
cognition · position
what children get right that current ai doesn't. an argument for using developmental cognition, not language fluency, as the target for what general intelligence should mean.
book and paper summaries.
a peek at what I'm currently thinking about. recent: yang & chen on neuromorphic intelligence, wegner on the illusion of conscious will, arendt on the banality of evil. earlier sequences run through ricoeur on time and memory, marcus on the painter soutine, and adorno's minima moralia in full.
loading recent episodes…
detailed industry case studies available on request.
email for accessopen to research collaborations, advising, speaking, and conversations.