Arthur Prat-Carrabin (He/Him)

Postdoctoral Research Scholar, Columbia University
  • United States of America

About Arthur Prat-Carrabin

On the job market!


I'm interested in how information is represented in the brain, from encoding-decoding mechanisms of perception to human inference under uncertainty. In my work, theory always informs experiments, and vice versa.

I am proud of this theoretical paper, presented at NeurIPS: it gives analytical approximations to the bias and the variance of the Bayesian mean. A related, experimental work on number perception is reported in this article, published in Nature Human Behaviour.
This Psychological Review paper looks at human inference in changing environments (change points), but with temporal structure in the signal. Finally, this work is a detailed investigation of human responses in a probabilistic inference task – and the results constitute quite a challenge to the Bayesian paradigm.

Feel free to reach out to talk about all these things!


Intro Content

Nature Human Behaviour

Our sense for numbers is imprecise, but efficiently so

In spite of their apparent limpidity, we are often imprecise in our representations of Arabic numerals. The brain is however "efficiently imprecise": more frequent numbers are perceived with less noise, and the brain adjusts this efficient coding, when the relative frequencies of numbers change.


Channels contributed to:

Behind the Paper


Online Elsewhere