For nearly fifty years, biophysicists believed that E. coli was an excellent chemical sensor. The theoretical limit on how accurately a small cell can measure a chemical concentration was established by Berg and Purcell in 1977: the fundamental constraint is the random arrival of signaling molecules at the cell's surface. Researchers measured E. coli's chemotactic performance and concluded it was operating close to this physical limit, maybe within a small factor.
Mattingly and colleagues measured the actual information rate (Nature Physics, 2026). E. coli encodes two orders of magnitude less information than the stochastic molecule arrival limit allows. The cells are not near-optimal sensors constrained by physics. They are mediocre sensors constrained by their own internal signal transduction machinery.
This is not a minor quantitative correction. The entire theoretical framework for bacterial chemosensing was built on the assumption that the external limit mattered. Researchers designed experiments to probe how cells approach the physical bound, developed mathematical models of optimal receptor design under external noise, and compared species by how closely they reached the diffusion limit. The organizing question of the field — how close can a cell get to perfect measurement? — assumed the answer was “very close.” The actual answer is “not remotely.”
What makes this structurally interesting is that the internal noise was always there. The signal transduction pathway in E. coli chemotaxis is one of the best-characterized molecular systems in biology. The receptors, the kinase cascade, the methylation adaptation — every component has been measured, modeled, and quantified. But the information-theoretic comparison between external and internal limits was not done until now. The measurements existed. The framework to compare them existed. The question was simply not asked in the right order.
A companion paper in the same issue found that the E. coli chemosensory array operates near a thermodynamic critical point, balancing signal gain against response speed through biologically tuned disorder. The system is not optimized for accuracy. It is optimized for a tradeoff that accuracy-focused models never considered. The critical point is a different organizing principle entirely — the field was measuring one performance axis while the system was optimized along another.
The pattern recurs. The cleaner wrasse mirror test, redesigned by Sogawa and colleagues (Scientific Reports, 2025), revealed that four-to-six-day latencies were measuring mirror habituation, not self-recognition development. When the protocol was reversed — mark first, mirror second — wrasse responded in 82 minutes. The cognitive bottleneck was never in the fish. It was in the experimental design.
In both cases, the external limit was assumed to be binding when the internal limit was actually binding. The consequence is not just a revised number. It is a revised framework. When you discover that the bottleneck is inside the system rather than at its boundary, every downstream analysis built on the external-limit assumption needs to be reconsidered. The field's questions were reasonable given the wrong premise. They are the wrong questions given the right one.