What if the way we think our brains filter information is actually an optical illusion of data?
For years, a central debate in neuroscience has centered on "neural geometry"—how our neurons organize complex information to make decisions.
The "Compression" Theory
A high-profile theory recently suggested that our brains use a "compression" mechanism, essentially acting as a gatekeeper that shrinks irrelevant details to make room for what matters. It is an elegant idea: when you are looking for a red car, your brain might physically compress "motion" data to focus on "color."
A New Technical Re-evaluation
A new analysis of macaque and human datasets suggests the story in the prefrontal cortex (PFC) is far more complex—and perhaps far less "compressed" than we thought. The controversy centers on the Frontal Eye Fields (FEF), a region critical for transforming visual information into eye movements.
Researchers took a second look at data from Rhesus macaques and humans to see if this "gating" was a biological reality or a mathematical ghost.
The Key Discovery
Researchers discovered that in the original macaque studies, certain variables were inherently linked:
- Motion direction was correlated with the target position.
- Color was not correlated.
The FEF is dominated by "choice" signals—the physical preparation to move the eyes. Therefore, the activity related to the upcoming movement was accidentally being read as an enhancement of "relevant" information.
Simulating a Different Mechanism
When the team applied simulations based on a "late-selection" mechanism—where the brain keeps all sensory data intact and only filters it at the very last moment of choice—they were able to reproduce the "compression" results without any actual neural compression occurring.
Why This Matters
This discovery challenges our understanding of brain efficiency. If the prefrontal cortex maintains "uncompressed" representations of irrelevant inputs, it means our brains are much more like a raw data recorder than a filtered Instagram feed.
We aren't deleting the noise early on; we are simply choosing to ignore it at the finish line.
The Unsolved Mystery: A Divergence Between Species
The analysis reveals a discrepancy:
- Human fMRI data (where inputs were uncorrelated) still showed signs of compression.
- Macaque single-unit recordings did not.
This could be due to a byproduct of the recording tools or could mean the FEF operates under different computational principles than the broader human prefrontal cortex.
The Current Conclusion
For now, the evidence suggests that the macaque brain prefers to keep its options open, maintaining a full, uncompressed view of the world until the very second it decides where to look.
Reference: “Are task representations gated in macaque prefrontal cortex?” by Flesch, T., Mante, V., Newsome, W., Saxe, A., Summerfield, C., and Sussillo, D. (Joint Correspondence regarding Flesch et al., 2022 and Mante et al., 2013).