Plural Intelligence

Science seeks to attain truths that hold for all individuals beyond their subjective perspectives. Formal sciences achieve this by deriving theorems within closed axiomatic systems without uncertainty. Empirical sciences based on data, however, must validate their propositions within open natural systems that contain regions hidden from our perception. Is it possible to reach truths (intersubjective agreements) in contexts of uncertainty? At the very least, we know the meaning of not lying (at least to oneself): not asserting or affirming more than what is known, while not concealing anything that is known. Mathematically, this is defined as maximizing uncertainty (or entropy) given the available information (or constraints), a principle known as MaxEnt [1].

Distributions adhering to the principle of not lying possess the property of intuitively aligning our individual subjectivities. For example, if we know there is a hidden gift behind one of three identical boxes, we naturally avoid assigning all our belief to a single box (since we lack absolute certainty about the gift’s location) and also avoid assigning greater belief to one box over the others (as we have no information to prefer one over another). Intuitively, all individuals, regardless of culture or ideology, ultimately acknowledge that the only possible belief distribution given the available information is the one that equally divides belief among the three boxes (maximizing uncertainty) without assigning belief outside of them (given the available information).

The principle of not lying constitutes the foundation upon which empirical sciences reach truths (intersubjective agreements) in contexts of uncertainty. The rules of probability, proposed in the late 18th century and since adopted as the reasoning system in all data-driven sciences, encode this principle. These rules are conceptually intuitive. The product rule (or conditional probability) updates belief distributions by preserving prior belief (prior information) that remains compatible with new information (data), thereby maximizing uncertainty given the available information. The sum rule (or marginal probability) predicts yet-to-be-observed events by incorporating all alternative (mutually exclusive) hypotheses, integrating all individual predictions.

In probability, hypotheses are evaluated through a sequence of predictions, where initial belief is filtered through surprise, the only source of information. A hypothesis that predicts with 1 (zero surprise) preserves all prior belief. A hypothesis that predicts with 0 (total surprise) becomes permanently false. Similarly, evolving life forms develop through sequences of survival and reproduction rates. A single zero in the sequence leads to the extinction of the life form. The multiplicative nature of hypothesis evaluation and life form selection processes is responsible for learning [2], both in probability and evolution. Not coincidentally, the standard model of evolution (replicator dynamic [3]) is structurally equivalent to Bayes' theorem [4].

Since, under multiplicative processes, the impact of losses is stronger than that of gains, the variants that thrive in probability and evolution are hypotheses or life forms that reduce fluctuations through plurality. This is evident in the evolution of our own life, which depends on at least four levels of cooperation with specialization, without which we could not survive: the cell with the mitochondrion, the multicellular organism, society, and the ecosystem [5]. Similarly, in probability, elementary hypotheses group together to form variables, variables relate to each other to form causal models, and systems of models form theories [6].

The advantage of plurality is not theoretical but practical. When plurality is disrupted in these systems, evident negative effects arise. In probability, arbitrarily selecting a single hypothesis from the space results in what is known as overfitting [7]. In evolution, the lack of genetic diversity leads to negative consequences known as inbreeding depression. In human history, the massive loss of cultural diversity caused during colonial-modernity by the imposition of a single type of society has led to increasingly evident environmental consequences [8]. Despite all advancements, metropolitan science remains unable to compensate for the loss of millennia-old knowledge, and the current ecological crisis continues to deepen [9].

Although no better reasoning system for uncertain contexts has been proposed in practical terms, the strict application of probability rules (the Bayesian approach) has historically been limited due to the computational cost required to evaluate the entire hypothesis space. While many models were solved completely until the late 19th century, especially in statistical physics, the 20th century saw the introduction of criteria that, to avoid computational costs, arbitrarily selected a single hypothesis from the space. It was not until the dawn of the 21st century that it became generally possible to compute optimal belief distributions given the available information across all scientific fields.

Causal arguments that correspond to the underlying causal reality are necessarily the higher-level hypotheses (theories) that generate the least surprise, and no artificial intelligence model, no matter how complex, can improve upon their performance. However, we are not always capable of formulating these causal arguments, nor are we always able to (approximate) the strict application of probability rules to evaluate them correctly. Although deep neural networks are trained by selecting a single parameter from the possible parameter space, recent advances have managed to mitigate negative consequences (overfitting) through a form of plurality similar to that employed by life in evolution, based on the coexistence of a vast number of individual units (neurons), leading to the emergence of true artificial intelligence (double descent) [12].

Just as all intelligence emerges from plurality, societies adapt to life through the coexistence of their local diversities. The accumulated experience of the world's most diverse communities has independently led to a universal obligation to give and receive, and to the development of reciprocity technologies that reactivate community bonds through exchange rituals (festive or coercive). Human institutions capable of managing common goods are of a communal type, where cycles of exchange with ecological systems are locally and directly regulated [13, 14]. For these reasons, our main goal is to promote Plural Intelligence among the Global South.

Referencias:

[1] Jaynes ET. Information theory and statistical mechanics. Physical review. 1957;106(4):620.
[2] Kelly, JL. A new interpretation of information rate. The Bell System Technical Journal. 1956.
[3] Taylor PD, Jonker LB. Evolutionary stable strategies and game dynamics. Mathematical biosciences. 1978;40(1-2):145–156.
[4] Czégel D, Giaffar H, Tenenbaum JB, Szathmáry E. Bayes and Darwin: How replicator populations implement Bayesian computations. BioEssays. 2022; p. 2100255.
[5] Maynard Smith J, Szathmary E. The Major Transitions in Evolution. New York: Oxford University Press; 1995.
[6] Winn J. Causality with gates. In: Artificial Intelligence and Statistics. Proceedings of Machine Learning Research; 2012. p. 1314–1322.
[7] Bishop CM. Pattern recognition and machine learning. Springer. 2006.
[8] Dussel E. Sistema mundo y transmodernidad. In: Modernidades coloniales. El Colegio de México México DF; 2004. p. 201–226.
[9] Segato, RL. La crítica de la colonialidad en ocho ensayos y una antropología por demanda. Prometeo (2013).
[12] Bishop, CM and Bishop, H. Deep learning: Foundations and concepts. Springer Nature. 2023.
[13] Ostrom E. Governing the commons: The evolution of institutions for collective action. Cambridge university press. 1990.
[14] Ostrom E. Beyond markets and states: polycentric governance of complex economic systems. American economic review. 2010.