Workshop at the Santa Fe Institute
Cowan Campus (1399 Hyde Park Road, Santa Fe NM 87501)
March 26-28, 2024
Meeting Summary
Traditionally, theory concerning reality has been conceived of as a human construct approximating some separate “physical reality.” However, several contemporary areas of research have begun to blur the boundary between our theories of reality and reality itself. One clear example of this is in the social sciences, where “social reality” and a model or theory of society, are often difficult to disentangle. For example, does bounded rationality game theory describe strategic interactions in markets, or does bounded rationality game theory define the “social reality” of markets? This disruption of our understanding of a clear distinction between theory and the reality that the theory describes extends far below the social sciences to encompass deep, and by now, widely held disciplinary positions, from the Many-Worlds interpretation of quantum mechanics, to the anthropic principle and cosmological inflation, to the “It from Bit” school of physics, to the idea of generalized observers in adaptive systems (from natural selection to cultural evolution). And these have been extended up the physical hierarchy generating more controversial (yet highly influential) frameworks, including Simulation Theory, The Free Energy Principle, the Principle of Computational Equivalence, and many theories of reflexivity and agency that bring us back toward the recursions of the social sciences. We seek to review these various frameworks and levels and to establish a quantitative basis (e.g. a parsimony or coherence principle) for relating them. Are there levels that have a natural rooting – what we declare as fundamental - which provides the most descriptive and predictive power if we formulate the other levels as its consequences? Is there one framework that should properly be viewed as providing the axioms from which all the others follow?
Meeting Description
The Two Realities: Material and Conceptual
We tend to think about reality in one of two ways. The first involves physically emergent hierarchies – building up from the most “fundamental” elementary particles, through associated “universal laws,” of physics, collective chemistry, adaptive matter, ecosystems, brains, mind, and ultimately, societies. In these hierarchies, there is no ontological distinction between the theory and the thing the theory describes / approximates.
The second describes conceptually emergent hierarchies – spanning logic, mathematics, natural language, natural science, and the arts. This perspective focuses on the cognitive and conceptual structures that humans create to describe the material hierarchies in which they are embedded. These hierarchies focus precisely on the distinction between the theory and the thing the theory describes / approximates – a distinction denied by the first type of hierarchy.
Much of this structure was adumbrated by Anderson in his More is Different paper/manifesto from 1972 and its sequels, which interwove these two ideas of emergence and associated hierarchies. Catching up with Anderson, several contemporary areas of research, integral or tangential to complexity science, have blurred the boundary between our theories of reality and reality itself.
One clear example of this is in the social sciences, where “social reality” and a model or theory of society, are often difficult to disentangle. For example, does bounded rationality game theory describe human strategic interactions, e.g., in markets? Or instead, does bounded rationality game theory constitute those strategic interactions, in the form of "laws" that govern them in those settings?
This disruption of our understanding of a clear distinction between theory and the reality that the theory describes extends far below the social sciences to encompass deep, and by now, widely held disciplinary positions, from the Many-Worlds interpretation of quantum mechanics, to the anthropic principle and cosmological inflation, to the “It from Bit” school of physics, to the idea of generalized observers in adaptive systems (from natural selection to cultural evolution). This disruption has also been extended up the physical hierarchy, generating more controversial (yet highly influential) frameworks, including Simulation Theory, The Free Energy Principle, the Principle of Computational Equivalence, and many theories of reflexivity and agency - which bring us full-circle back toward the recursions of the social sciences.
These new areas of research raise a rather troubling, and of course ancient, question. Should we conceive of physical and complex reality as the products of observers and minds working with natural language, mathematics, and computation? Or should we instead conceive of physical reality as separate from observers and minds, emerging from math, through physics, up to neurobiology, and then society, etc.? Most of us were trained as materialists, placing matter before mind. But this easy assumption has been repeatedly questioned and these assaults have become more frequent in recent years.
This blurring between theory and the reality described by the theory is not half as weird as it might at first sound. When one considers artificial life and artificial intelligence, logic and mathematics must precede in silico ontologies of any kind (i.e. it takes human-formulated natural science to engineer artificial life, and so forth). In these cases, we see very clearly that ideas need to precede materials.
This is of course the starting point of simulation theory that considers all of reality emergent from formal logic and computation. Once dismissed as science fiction, there are now research projects seeking to find testable predictions of simulation theory. Interestingly these tests of a simulated reality are not unlike Turing tests for discovering intelligent agents.
The Reality Ouroboros
We seek to review these various frameworks and establish a quantitative basis (e.g. a parsimony or coherence principle) for relating them, and maybe even deciding among them. Are there levels in the hierarchies that have a natural rooting which provides the most descriptive and predictive power in the levels that follow?
We might start by classifying all these frameworks in terms of the preferred “insertion points” or starting points of an arbitrarily fine-grained Ouroboros (which begins and ends at the same point) encoding successive emergent levels. The point where the mouth and the tail of the serpent intersect is what we shall declare to be fundamental:
Platonists, A-lifers, and Simulation Theorists, start at the insertion point of Mathematics (ideas). This assumes that the math that we favor needs to be understood in terms of brains that have evolved under physical constraints, which are themselves statements of mathematics. This approach takes the Ouroboros cycle to be best understood in terms of fundamental mathematical symmetries:
Math ➙ Physics ➙ Neurobiology ➙ Mind ➙ Math.
Biological materialists start at the insertion point of neurobiology, arguing that our models of the brain cannot be decoupled from the architecture and constraints of the brain itself:
Neurobiology ➙ Mind ➙ Math ➙ Physics ➙ Neurobiology.
Origin of life researchers start at the insertion point of Chemistry. They seek to identify crucial broken symmetries that propagate up through living matter and ultimately support the evolution of brains and minds whose models of physics and chemistry dictate how we think about life:
Chemistry ➙ Life ➙ Neurobiology ➙ Mind ➙ Math ➙ Physics ➙ Neurobiology ➙ Chemistry.
Research fields might be defined by the point at which they start in the Ouroboros cycle: the level that they declare to be fundamental for their investigations. Physics likes to start with particles and fields and see how far simple symmetries might be explored to explain matter. Biologists like to start with organic chemistry and broken symmetries to determine how far this might explain functional organization. And psychologists like to start with mind and ask about the origins of mathematics upon which any physical theory will ultimately be constructed.
Following the pioneering ideas of Phillip Anderson mentioned above, it might be assumed that there is an arbitrary choice of starting point: no one science is more fundamental than any other. Once having arbitrarily rooted the Ouroboros, researchers explore the implications of this choice along the emergent hierarchy. This is in one telling the subject of complexity science – surveying the many distinct approaches to reality that follow from our assumptions about what constitutes the most fundamental level of explanation, the many effective theories we can use to root our analyses.
We believe that the implications of these questions are rather profound and touch on fundamental questions in many active fields including complexity-related research. These include:
- the foundations of physics, neurobiology, and cognitive science;
- theory of logic, metamathematics, and category theory;
- computational complexity, impossibility results of Turing machines, dynamical systems theory;
- ontic structural realism, epistemology (e.g., epistemic logic), ontology
Complexity, Emergence, and Reality
We have arrived at a rather exciting time in science where philosophical questions have once again become centrally important.
At SFI, several recent Fractal Faculty have been exploring the four questions above: Sean Carroll, recently appointed as Professor of Physics and Philosophy at John's Hopkins; Kyle Harper (ontology of disease); Niall Ferguson (epistemology of collective dynamics) investigating the natural scientific foundations of history; Venki Ramakrishnan writing about the multi-scale properties of death -- molecules to ethics; and Carlo Rovelli, cosmologist focusing on quantum gravity and the thermodynamic arrow of time. And numerous recent papers by SFI resident faculty touch on these issues. These include our own work, as well as the work of Chris Kempes, Jessica Flack, and Geoffrey West.
We argue that few other institutions besides SFI are sufficiently eclectic and multidisciplinary to convene people from all these fields with the mandate to look at the entire Ouroboros, considering it as a whole. In part, the role of SFI in modern science is to initiate and pursue these kinds of unifying projects that might formerly (with reasonable cause at that time) have been declared tangential to the progress of science – but no longer. As we have argued, this is in particular the case for the many aspects of the Ouroboros. The Many Worlds perspective with its crucial observer dependence is not marginal but mainstream. The same case can be made for numerous observer-first frameworks from natural selection to game theory. And recent interest in machine learning and simulation theory have made the issues even more fascinating.
Lastly on an institutional note we see this meeting as a return to the higher risk, more expansive style of science at SFI. The community that it will attract will be broader, and the ideas both challenging and provocative. Complexity science need not mature at the expense of a little healthy madness.
Outcomes
Thomas Kuhn famously wrote of periods of normal science - puzzle solving phases - and revolutionary science - phases involving empirical anomalies – and, as importantly, changes in the “disciplinary matrix” that define fields. All revolutions question the balance of ideas that are deployed to investigate reality. In this meeting we are pursuing the idea that: (1) the current disciplinary matrix, that is somewhat inadequate to address; and (2) the anomalies of theory or model-based mechanisms that pervade complex systems. The most important, and obviously highly optimistic, outcome of this meeting would be starting a far-ranging reconfiguration of our current disciplinary-matrices, allowing for fields that are currently unconnected in their practice to extend collaborative connections. A more proximate outcome is to promote awareness of shared frameworks and methodologies that relate to our central question -- the nature of human theories and their role in understanding systems that theorize. This strikes us a rather important intellectual and practical outcome in a world where rules and algorithms are being built into numerous decision-making processes. And the approximately recursive structures that emerge out of “outsourcing” values and procedures are in need of new forms of understanding.
Contact
Email: rtursi@santafe.edu