Empirical knowledge is life

The reciprocal coexistence of diversities

The current massive loss of biodiversity was preceded by a massive loss of cultural diversity. Despite all the advances, metropolitan science was not able to compensate for the loss of millenarian knowledge caused by colonial-modernity. Evolution, like Bayesian inference, needs diversity to function properly. Long-term coexistence with ecological systems implies recovering the reciprocal coexistence between autonomous indigenous communities. Because, in short, empirical knowledge is life.


In the last third of the history of the Universe, a form of organization of matter capable of self-replication appeared on Earth. The growth of these lineages followed multiplicative and noisy processes: a sequence of survival and reproductive probabilities. In it, the impacts of losses are often stronger than those of gains, if there is even one zero in the sequence, we are extinct. Because variance reduction is really important, cooperation has a fundamental evolutionary advantage. This caused life to acquire an extraordinary complexity throughout its history.

The skills that allow us to create complex cultures developed due to the prior emergence of cooperative breeding, which produced an environment that favored the selection of offspring with special capacities for mutual understanding. The ability to transmit knowledge from generation to generation produced radical consequences for our species. Before the cultural transition, we were in serious danger of extinction, as evidenced by the low diversity of the human genome. After the cultural transition, our species was able to occupy all the ecological niches of the earth, as no other terrestrial vertebrate had ever done before.

Science is a human institution that aims to formulate universally valid propositions, both interculturally and intersubjectively. Formal sciences validate them by means of theorems, results derived from the internal rules of a closed axiomatic system. Empirical sciences, in contrast, must validate them within open systems, which always involve a degree of uncertainty. Probability theory is by far the most widely used approach to represent uncertainty, and it is already remarkable that it has been derived from a large number of different axiomatic systems.

Probability theory only has two rules. The sum rule imposes a cooperative pact among the hypotheses of the same model: predictions are made with the contribution of all of them. The product rule is the multiplicative process by which hypotheses (and models) are selected, the same process by which evolution selects among different strategies, the reason why cooperation has an evolutionary advantage. Together, they update beliefs by maximizing uncertainty given empirical (data) and formal (causal models) information, which guarantees the validity of empirical knowledge. Why?

The “principle of indifference” is the cornerstone of the empirical sciences, an epistemological criterion of intercultural validity, present in all the peoples of the planet. We will all agree that, in the absence of prior information, the only correct way to distribute belief among a (finit) set of hypotheses is in equal parts (details at Honest Beliefs). Then, the only universally valid propositions for empirical science are those that maximize diversity (i.e. uncertainty).

The advantage of cooperation is not principled, it is pragmatic. Due to the multiplicative nature of the selection processes, both evolutionary and probabilistic, the breakdown of the cooperative pact negatively affects those who promote it without the need to introduce punishments. The overfitting problem associated with frequentist techniques is a direct consequence of selecting a single hypothesis. The problem of ecological crisis associated with colonial-modern society is a direct consequence of imposing a single type of society.

The solution to the ecological crisis (and overfitting) consists in recovering the coexistence between the various forms of life (or hypotheses). Evolution is Bayesian inference for the real world, and for it to work properly, diversity is required. Ultimately, empirical knowledge is life.

Avatar
Gustavo Landfried
Bayesian Data Scientist

Empirical knowledge emerges as life does