Michael J. Berry II

Michael J. Berry II is a Princeton neuroscientist whose lab studies the retina not as a simple camera but as a computing organ. Search results give you three facts about him — professor, retina recordings, ex-physicist — and then run dry. None of them tell you that the young graduate student who found the lab’s measurement methods so inadequate that he built a better sensor was Dario Amodei, the future Anthropic CEO. That is where this profile starts, because it is the most human doorway into what the Berry Lab actually does (Big Technology).

The Amodei chapter

Amodei arrived at Princeton carrying grief — his father Riccardo had died in 2006 — and redirected his graduate work from theoretical physics toward biology. In Berry’s retina lab, he was, in Alex Kantrowitz’s reporting, “so dissatisfied with the methods available to measure the retina’s signals that he co-invented a new, better sensor to pick up more data.” The lab considered this atypical: impressive and nonconformist. His dissertation won the Hertz Thesis Prize, given for academic work with real-world applications. And Berry told Kantrowitz plainly that Amodei was “the most talented graduate student he ever had” — though, in the same breath, he noted that Amodei’s drive toward technological progress and teamwork “didn’t play well in a system built to recognize individual achievement” (Big Technology).

That story is a neat encapsulation of Berry’s whole project. His lab’s central question is how thousands of neurons connected in local microcircuits “combine to carry out novel and powerful computations,” and it studies that question in the retina — the thin sheet of tissue at the back of the eye that encodes visual information for the brain (Princeton University (Berry Lab site)). A student who wanted better tools for listening to those computations was, in a sense, working on the lab’s most characteristic problem.

From particle physics to retinal circuits

Berry spent years training as an experimental physicist before he ever recorded a neuron. He studied physics and philosophy at UC Berkeley, earned a Ph.D. in physics at Harvard, and then stayed on for another five years of postdoctoral research — experiments probing the properties of electrons — before his strong theoretical bent began pulling him elsewhere. The turning point came when he started applying his mathematics to neuroscience problems, such as how retinal cells anticipate the future position of a moving object. That shift, in his words, let him return to his first love: “Biology was my first love,” he told the Princeton Weekly Bulletin in 2007. “Computational neuroscience is a hybrid field, one where I can study biological behavior with the tools I learned in physics” (Princeton University).

He joined Princeton’s faculty in 1999, in the Department of Molecular Biology, drawn by a neuroscience institute with no large medical school to skew its priorities — a place where “conceptual and mathematical ideas” could shape the work. Tenure followed, a few months before the 2007 profile. His pitch for studying the eye is characteristically brisk: retinal cells are closely related to neurons elsewhere in the brain, “so studying retinal circuits can reveal processes common throughout the entire brain” (Princeton University).

What the recordings showed

The Berry Lab’s signature technical achievement is density. Using a dense grid of electrodes and an algorithm that identifies ganglion-cell signals arriving on multiple electrodes, the lab recorded from all of the ganglion cells in a patch of the retina — “a feat that has not been achieved in any other neural circuit” at the time, according to Pew’s profile of his 2000 Biomedical Scholars award. Under natural visual conditions, the cells fired infrequent bursts of spikes with millisecond timing precision and were otherwise completely silent (The Pew Charitable Trusts).

Princeton’s own research office later credited Berry with recordings “covering the greatest number of neurons per unit area ever measured in the retina,” plus “superior algorithms for identifying the activity of these neurons” that other researchers were likely to adopt (Princeton University (Office of the Dean for Research)).

The findings mattered because they showed single-neuron thinking was insufficient. The 2007 profile describes the core insight: brain cells behave in “rich and interesting ways as groups,” and you would never see it without sophisticated statistical analysis. Noise that would corrupt one cell’s message — one misfire that says “I see a rabbit running” when there is really a ball bouncing — is overcome because nearby cells fire together in collective patterns whose meaning cannot be decomposed into individual contributions. That group-level code, it turned out, is error-correcting, and its constituent neurons exhibit “qualitatively different receptive field properties” as codewords than they do alone (Princeton University; Princeton Neuroscience Institute).

A favorite result concerned prediction. The retina, the lab found, carries out “a diverse set of surprisingly sophisticated computations,” of which the most arresting is the ability to make simple predictions about what happens next in the visual input. When a moving object suddenly reverses, a group of retinal cells fires in a unique pattern — signaling, to the brain, a violation of the retina’s prediction. Berry, the 2002 New Jersey state squash champion, said he thought about athletics when thinking about retinal work: “Circuits in the retina itself have to make lots of predictions regarding movement, and what happens if those predictions are violated?” (Princeton University; Princeton Neuroscience Institute).

More recently the lab has pushed beyond the retina, using optical imaging in mouse primary visual cortex — in collaboration with the Tank lab — to find that neurons in cortical layers 2/3 carry out a broader class of predictive computations in response to repeated temporal sequences, feeding “novel hypotheses about the different computations carried out within each layer of the neocortex” (Princeton Neuroscience Institute).

What this profile does not claim

• It does not give an exact birth date or birthplace; no source read in full provides them.
• It does not describe the mechanism of the retinal sensor Amodei co-invented; the reporting consulted describes the dissatisfaction and the invention in general terms only.
• It does not claim Berry still heads the lab in 2026; it reports that Princeton’s own sites currently list him as its head and records that no contrary evidence was found in the read sources.
• It does not adjudicate which title wording is current (“Professor of Neuroscience” vs. the older “Associate Professor of Molecular Biology”); each source’s own wording is preserved where used.