Are biological mechanisms optimized, or do they function poorly, evidence of their “poor design”? William Bialek, professor of physics at Princeton, has discovered that assuming organisms use optimized mechanisms for accomplishing their most crucial tasks is quantitatively testable. And at least in multiple tested cases, the assumption of optimality turns out to be correct. 

The Aesthetic of Optimization

But assuming optimality worries many scientists. Bialek notes, “optimization comes along with an aesthetic that you might or might not find appealing.” (Bialek 2024) He writes, “For some, optimization is obvious because evolution has had billions of years to get things right. For others, optimization is nonsense because evolution is not about being best, but about being better than the competition.” It seems what Bialek is alluding to here is that highly optimized systems have not historically been a prediction of Darwinian evolution, but rather of intelligent design theory. 

In the world of physics, the expectation is that a wide range of phenomena can be explained using a small number of principles. Indeed, the motion of the sun and moon, even weather patterns, are understood in this manner. But surely the same cannot be said of life! Or can it? Life, after all, is different. It self-organizes with stunning accuracy and exhibits diversity beyond our ability to imagine. As a physicist, Bialek asks: Is there something about the diversity of life that can be explained with a small set of general principles? What are the theoretical principles of biology?

To Build a Fruit Fly 

In his recent paper, “Ambitions for theory in the physics of life,” Bialek outlines a theoretical principle of life as the optimization of information flow. As an illustration he offers the development of a fruit fly from a single cell to a multicellular organism.

Just three hours after an egg is laid, the blueprint for the animal’s segments can be observed through fluorescence microscopy of particular proteins. These proteins, which form distinct stripes, are the products of just eight genes, and the synthesis of these proteins is controlled by regulatory genes known as gap genes.

When the mother lays an egg, she provides three key information-containing inputs to kick off the exquisite segmentation patterning in the embryo. These inputs have a simple distribution: one is high at one pole of the egg, another is high at the opposite pole, and the third is high in the middle. These maternal inputs activate the gap genes, which in turn co-regulate each other. The gap genes then regulate the pair-rule genes, which establish the blueprint for the insect’s segmented body. Thus, to build a fruit fly, information flows from the maternal inputs, through the gap gene network, and finally to the pair-rule genes that produce the striped pattern. This information flow is near the theoretical optimum.

Bialek has taken these biological observations and fluorescent images and translated them into mathematical equations. But how exactly did he and his team do this? Physicists analyze biology using mathematical concepts to model biological phenomena. For example, they might model the synthesis of gene B, which depends on gene A, using an equation that describes the rate of change of a variable such as protein concentration over time, where the first term in the equation represents the rate of synthesis and the second term, which is subtracted from the first, represents the decay.

Remarkable Precision

The outcome of Bialek’s elegant mathematical descriptions is that cells identify their position in the developing fly embryo with remarkable precision — accuracy to within 1 percent — using the concentration of the gap proteins. Consider gap protein 2 which has a high concentration in the middle of the embryo. Cells in this region experience the maximum concentration of gap protein 2 and thus “know” they are in the middle. This protein provides precise information at the center but has ambiguity at the flanks and uncertainty at the ends. If this were the only gap protein, cells at the head and tail of the embryo would be indistinguishable. However, the presence of multiple gap proteins removes this ambiguity, enabling cells to determine their position. For this to happen, cells must read out the information in the gap protein concentration optimally.

Thus, not only is the measuring of the concentration optimal, but concentration of the inputs appears optimized as well. Bialek hypothesizes that the concentration of maternal inputs should be centered on the point of maximum sensitivity as well as being distributed far enough from each other that they can drive the production of the gap proteins through the complete dynamic range. 

Bialek notes that the best way to achieve this is to “use inputs in inverse proportion to their noise levels.” By noise, he means the fluctuation in concentration around the mean. Bialek says this is exactly how we communicate — i.e., we avoid using words we can’t spell or, when speaking a foreign language, we steer clear of grammar that we might use incorrectly. Bialek reports that this is what occurs in the developing fly embryo: “Thus optimization of information transmission predicts a match between the positional noise levels and the distribution of cell positions, and this match is sufficiently good that it brings the embryo within 2 percent of the optimum.” Indeed, after analysis, the concentrations of the inputs appear optimized: “Somehow the signal and noise (and even the covariances) of all four gap genes conspire to generate nearly constant precision.”

In Summary

The optimization of information flow in biological systems (meaning the information flow from the maternal inputs, through the gap gene network and then the pair-rule genes, resulting in the cell’s capacity to measure its position at the 1 percent accuracy level) is not just a fascinating area of study. It’s also a powerful demonstration of how assuming optimality — the best design possible — can provide deep insights into the complexity of life. 

Bialek’s findings have implications for theories about origins, namely evolution and design. For example, Bialek’s demonstration that optimization is present in fruit fly development requires evolutionary biology to demonstrate that the mechanism of random mutation and natural selection can push to such optima. Bialek himself surmises that this is the case. But strong evidence already exists that evolution neither has the time nor the creativity to explore the degrees of freedom necessary to arrive at such optima (here, here, here).

Design and engineering, on the other hand, present a compelling alternative. Designers, by their nature, operate with foresight and intent, qualities that seem necessary for achieving optimized systems. And we know optimality is something designers can achieve, because they operate outside the system with a greater understanding of the goals and with the ability to use mathematical abstractions. Ultimately, the scientific community must remain open to whichever theory best aligns with the data, allowing evidence, rather than preconceptions, to guide the debate. I hope Bialek’s work challenges us all to reconsider the mechanisms behind the remarkable optimality observed in biological systems.

References

• Bialek, William. 2024. “Ambitions for Theory in the Physics of Life.” SciPost Physics Lecture Notes, no. 84 (August). https://doi.org/10.21468/scipostphyslectnotes.84.