Tal Golan

This account is verified. Learn more

@TalGolanNeuro

Assistant professor @ Ben-Gurion University. Studies and tweets about human and machine vision.

Israel brainsandmachines.org Joined July 2016

900 Following 1.1K Followers 784 Posts

Latest posts 5

Pinned post

Tal Golan @TalGolanNeuro Aug 28

How can we design experiments that make models disagree? One section of our new @NatRevNeurosci Review with @KriegeskorteLab and @SchuettHeiko examines studies that used stimulus sets designed to elicit distinct predictions from competing models. Full text link at the end. 1/14

Tal Golan @TalGolanNeuro Nov 4

Creative usage of diffusion to parametrize experimental stimuli, "fading" between non-faces and faces.

Hamid Karimi-Rouzbahani @HamidRouzbahani Nov 3

What do you see in these images? 👁️💬

Faces appearing in everyday objects (face pareidolia) offers a unique window into visual perception. But does it start with bottom-up sensory extraction or top-down expectation?

Our work addresses this debate doi.org/10.64898/2026.…! 👇🧵

View quoted post on X

Tal Golan @TalGolanNeuro Oct 31

Some more AI-assisted teaching. This week, I gave our @bengurionu Cognitive and Brain Sciences sophomores a primer in multivariable calculus to help them develop an intuitive understanding of gradient descent, building on a smart lecture plan originally developed by @OrenShriki.

I used Codex to develop an interactive, fully client-side @plotlygraphs demo that visualizes how fixing all but one variable collapses a multivariable function into a function of a single variable—and how partial derivatives are simply ordinary derivatives of these one-variable functions.

Because the demo runs entirely in the browser, students can open it themselves and interactively explore the surfaces, slices, and derivatives.

The demo is available online and for download here:
brainsandmachines.org/demos/partial-…

0:59

Tal Golan @TalGolanNeuro Oct 13

From a NeuroAI perspective, it's really annoying that feedforward neural nets rule the world. Unrolling equivalence notwithstanding, what inductive biases does recurrence buy us? Looking forward to reading this.

Amirhossein Farzmahdi @farzmahdi Oct 13

1/5  New preprint: what does lateral recurrence add to visual representations?
Recurrence reuses the same weights over time: more computation, no new parameters.
Does that extra computation refine representations? If so, which ones?
doi.org/10.64898/2026.…

View quoted post on X

Tal Golan @TalGolanNeuro Oct 3

Interesting thread.

Shahab Bakhtiari Verified account @ShahabBakht Oct 3

I had a feeling recently that seems to be corroborated by chats with colleagues, so I decided to share it here.

A lot of systems neuroscientists driven by algorithmic-level questions are going through what looks and feels like a research depression. 🧵

(1/n)
View quoted post on X