Finally, we noticed orthographic stimuli (symbols, icons, multi-object ensembles) loading strongly onto some shared axes. Follow-up analyses confirmed a shared neural signature for this content, consistent with precursor mechanisms in macaques for human orthographic rep's.

Factorizing this cross-species space, we find interpretable dimensions, many of which combine visual features and concept-level properties. E.g., factors 6 and 7 both capture animals, but distinguish right- vs left-facing pose. Next up, we also find several key differences:

We find that macaque and human IT share a surprisingly rich geometry of object space. In total, we recover around 90 reliably shared axes, many of which show interpretable clusters of content. Even the first two axes already contain a lot of richness:

We analyzed human fMRI and macaque multi-unit IT responses to the same 8640 naturalistic object images. Specifically, we built a framework for cross-species alignment that seeks to discern similarities and differences without specifying beforehand what that might look like.

Are you attending #VSS2026? Come check out my talk on cross-species alignment for finding shared and distinct representational geometries in primate IT. Saturday, May 16, 2026, 3.30pm, Talk Room 1

5/ We also share some simulations and proofs to illustrate the core point. Overall, our aim is to make the computation of noise ceilings more consistent across future work, and we share several tips for achieving this.

4/ To this end, we offer a basic intuition with math & visuals to explain how reliability maps onto model performance. In a nutshell, split halves both contain measurement noise, but a True model does not; so the former is doubly attenuated, causing ceiling underestimation.

3/ We analyzed the literature and found that about 60% of the sampled literature uses a mapping that makes models appear closer to ceiling than intended. The goal of this paper is to show the statistical underpinnings of why the above mappings follow.

Our first point: this distinction collides with other accounts in the literature. We catalogue some of the diverse meanings and practices associated with "bottom-up" and "top-down" neuroscience.

The critiqued paper outlines two research cultures: A bottom-up, precision-first, approach that emphasizes control and iterative steps. And a top-down, accuracy-first approach that values coarse-grained analysis & tackling big questions head-on. www.nature.com/articles/s41...

On the whole, we view our contribution to be one of clarification. If two scientists have starkly different conceptions in mind when they discuss oscillations in their writing or conversation, this will hamper empirical progress.

Finally, we outline a view in which oscillations in processes across levels of organization orchestrate neuronal computation via three processing syntaxes. We explain and consider empirical results in favor of each mode.

Next, we turn to oscillations specifically. There can be oscillations in processes and measurements, and one may happen without the other. We spell out criteria for inferring oscillations-in-process, and we argue oscillatory e-fields have special causal roles.

To get our analysis going, we introduce two distinctions. First, we delineate measurements and processes. Second, we separate causal and inferential relevance. We take these axes and analyse the relations between field potentials, electric fields, neurobiological processes, etc.

PhD thesis submitted... Now on to Germany for a postdoc! 🇩🇪