Complexity Thoughts: Issue #59
Unraveling complexity: building knowledge, one paper at a time
If you find Complexity Thoughts interesting, follow us! Click on the Like button, leave a comment, repost on Substack or share this post. It is the only feedback I can have for this free service. The frequency and quality of this newsletter relies on social interactions. Thank you!
→ Don’t miss the podcast version of this post: click on “Spotify/Apple Podcast” above!
Foundations of network science and complex systems
Decomposing causality into its synergistic, unique, and redundant components
This new method (named SURD) leverages information theory to decompose causality into unique, synergistic and redundant contributions, while quantifying causal leakage. Unlike traditional pairwise approaches, it simultaneously analyzes all time-series variables to build a causality map that distinguishes exclusive, conjunctive and overlapping influences. From the benchmark tests on predator–prey dynamics, climatological data and multispecies marine systems, it seems that it is able to recover complex causal architectures where existing methods fail. Its “model-agnostic” framework enables application across multiple domains: but to which extent can it be applied to networked systems with many units? We’ll see!
The paper calls for “model-free approaches for causal discovery” and mentions convergent cross-mapping: we disagree on the choice of the label, because there is nothing really model-free, as I have recently discussed in two short essays (the second one focusing exactly on CCM):
I perfectly understand that using “model-free” for information-theoretic measure is common, but still can’t really get used to it.
Causality lies at the heart of scientific inquiry, serving as the fundamental basis for understanding interactions among variables in physical systems. Despite its central role, current methods for causal inference face significant challenges due to nonlinear dependencies, stochastic interactions, self-causation, collider effects, and influences from exogenous factors, among others. While existing methods can effectively address some of these challenges, no single approach has successfully integrated all these aspects. Here, we address these challenges with SURD: Synergistic-Unique-Redundant Decomposition of causality. SURD quantifies causality as the increments of redundant, unique, and synergistic information gained about future events from past observations. The formulation is non-intrusive and applicable to both computational and experimental investigations, even when samples are scarce. We benchmark SURD in scenarios that pose significant challenges for causal inference and demonstrate that it offers a more reliable quantification of causality compared to previous methods.
Data-driven causal analysis of observational biological time series
Although this review is tailored to computational biology, I think it’s about methods of general interest (ie, foundational).
Complex systems are challenging to understand, especially when they defy manipulative experiments for practical or ethical reasons. Several fields have developed parallel approaches to infer causal relations from observational time series. Yet, these methods are easy to misunderstand and often controversial. Here, we provide an accessible and critical review of three statistical causal discovery approaches (pairwise correlation, Granger causality, and state space reconstruction), using examples inspired by ecological processes. For each approach, we ask what it tests for, what causal statement it might imply, and when it could lead us astray. We devise new ways of visualizing key concepts, describe some novel pathologies of existing methods, and point out how so-called ‘model-free’ causality tests are not assumption-free. We hope that our synthesis will facilitate thoughtful application of methods, promote communication across different fields, and encourage explicit statements of assumptions. A video walkthrough is available:
Spectral Dimension Reduction of Complex Dynamical Networks
Dynamical networks are powerful tools for modeling a broad range of complex systems, including financial markets, brains, and ecosystems. They encode how the basic elements (nodes) of these systems interact altogether (via links) and evolve (nodes’ dynamics). Despite substantial progress, little is known about why some subtle changes in the network structure, at the so-called critical points, can provoke drastic shifts in its dynamics. We tackle this challenging problem by introducing a method that reduces any network to a simplified low-dimensional version. It can then be used to describe the collective dynamics of the original system. This dimension reduction method relies on spectral graph theory and, more specifically, on the dominant eigenvalues and eigenvectors of the network adjacency matrix. Contrary to previous approaches, our method is able to predict the multiple activation of modular networks as well as the critical points of random networks with arbitrary degree distributions. Our results are of both fundamental and practical interest, as they offer a novel framework to relate the structure of networks to their dynamics and to study the resilience of complex systems.
Evolution
Footprint evidence for locomotor diversity and shared habitats among early Pleistocene hominins
I know that this is not the best section for this paper, but I didn’t ever plan to have one about anthropology. Nevertheless, I think this paper is very interesting for two reasons:
it shows how science can discover incredible events from limited data
it supports the possibility that our evolutionary history is more complex than the usually represented linear timeline
It is now well accepted that hominin evolution is a story of many lineages existing contemporaneously. Evidence for this pattern has mostly come from fossils being dated to similar time periods. Hatala et al. describe hominid footprints from 1.5 million years ago in the Turkana Basin in Kenya that were made by two different species within hours or days of each other (see the Perspective by Harcourt-Smith). Analyses showed that the footprints were made by individuals with different gaits and stances, and the authors hypothesize these to be Homo erectus and Paranthropus boilei. Although fossils of both species occur in the area, these footprints show that they coexisted and likely interacted. —Sacha Vignieri
For much of the Pliocene and Pleistocene, multiple hominin species coexisted in the same regions of eastern and southern Africa. Due to the limitations of the skeletal fossil record, questions regarding their interspecific interactions remain unanswered. We report the discovery of footprints (~1.5 million years old) from Koobi Fora, Kenya, that provide the first evidence of two different patterns of Pleistocene hominin bipedalism appearing on the same footprint surface. New analyses show that this is observed repeatedly across multiple contemporaneous sites in the eastern Turkana Basin. These data indicate a sympatric relationship between Homo erectus and Paranthropus boisei, suggesting that lake margin habitats were important to both species and highlighting the possible influence of varying levels of coexistence, competition, and niche partitioning in human evolution.
Ecosystems
A clear longitudinal gradient in species richness across oceans is observed in extant marine fishes, with the Indo-Pacific exhibiting the greatest diversity. Three non-mutually-exclusive evolutionary hypotheses have been proposed to explain this diversity gradient: time for speciation, center of accumulation, and in situ diversification rates. Using the morphologically disparate syngnatharians (seahorses, dragonets, goatfishes, and relatives) as a study system, we tested these hypotheses and additionally assessed whether patterns of morphological diversity are congruent with species richness patterns. We used well-sampled phylogenies and a suite of phylogenetic comparative methods (including a novel phylogenetically corrected Kruskal-Wallis test) that account for various sources of uncertainty to estimate rates of lineage diversification and morphological disparity within all three major oceanic realms (Indo-Pacific, Atlantic, and eastern Pacific), as well as within the Indo-Pacific region. We find similar lineage diversification rates across regions, indicating that increased syngnatharian diversity in the Indo-Pacific is due to earlier colonizations from the Tethys Sea followed by in situ speciation and more frequent colonizations during the Miocene coinciding with the formation of coral reefs. These results support both time for speciation and center of accumulation hypotheses. Unlike species richness unevenness, shape disparity and evolutionary rates are similar across oceans because of the early origin of major body plans and their subsequent spread via colonization rather than in situ evolution. Our results illustrate how species richness patterns became decoupled from morphological disparity patterns during the formation of a major biodiversity hot spot.
Neuroscience
Multiscale organization of neuronal activity unifies scale-dependent theories of brain function
Conserved multiscale brain organization across five phylogenetically diverse species
Unifies scale-dependent theories of brain function, balancing efficiency and resiliency
Hierarchical structure permits multiple timescales and enhances information flow
Enables flexible cross-scale reconfigurations of neural activity during behavior
Brain recordings collected at different resolutions support distinct signatures of neural coding, leading to scale-dependent theories of brain function. Here, we show that these disparate signatures emerge from a heavy-tailed, multiscale functional organization of neuronal activity observed across calcium-imaging recordings collected from the whole brains of zebrafish and C. elegans as well as from sensory regions in Drosophila, mice, and macaques. Network simulations demonstrate that this conserved hierarchical structure enhances information processing. Finally, we find that this organization is maintained despite significant cross-scale reconfiguration of cellular coordination during behavior. Our findings suggest that this nonlinear organization of neuronal activity is a universal principle conserved for its ability to adaptively link behavior to neural dynamics across multiple spatiotemporal scales while balancing functional resiliency and information processing efficiency.
Distinct synaptic plasticity rules operate across dendritic compartments in vivo during learning
Synaptic plasticity underlies learning by modifying specific synaptic inputs to reshape neural activity and behavior. However, the rules governing which synapses will undergo different forms of plasticity in vivo during learning and whether these rules are uniform within individual neurons remain unclear. Using in vivo longitudinal imaging with single-synapse resolution in the mouse motor cortex during motor learning, we found that apical and basal dendrites of layer 2/3 (L2/3) pyramidal neurons showed distinct activity-dependent synaptic plasticity rules. The strengthening of apical and of basal synapses is predicted by local coactivity with nearby synapses and activity coincident with postsynaptic action potentials, respectively. Blocking postsynaptic spiking diminished basal synaptic potentiation without affecting apical plasticity. Thus, individual neurons use multiple activity-dependent plasticity rules in a compartment-specific manner in vivo during learning.
Mapping global brain reconfigurations following local targeted manipulations
“our study underscores the complex consequences of focal brain region perturbations, providing insights into brain function and the implications of localized modulation”
Specific changes in one region can influence the activity throughout the entire brain, a phenomenon known as diaschisis. This study combines advanced imaging techniques and personalized brain simulations in mice to investigate how targeted brain interventions, such as lesions or temporary silencing of certain regions, reshape global brain connectivity. The findings provide insights into why some interventions lead to reduced connectivity (hypoconnectivity) while others result in increased connectivity (hyperconnectivity), as observed in the literature. By elucidating the mechanisms underlying diaschisis, this work establishes a framework for understanding the widespread effects of localized brain injuries or interventions and for developing more precise therapeutic strategies that address brain dynamics across multiple scales.
Understanding how localized brain interventions influence whole-brain dynamics is essential for deciphering neural function and designing therapeutic strategies. Using longitudinal functional MRI datasets collected from mice, we investigated the effects of focal interventions, such as thalamic lesions and chemogenetic silencing of cortical hubs. We found that these local manipulations disrupted the brain’s ability to sustain network-wide activity, leading to global functional connectivity (FC) reconfigurations. Personalized mouse brain simulations based on experimental data revealed that alterations in local excitability modulate firing rates and frequency content across distributed brain regions, driving these FC changes. Notably, the topography of the affected brain regions depended on the intervention site, serving as distinctive signatures of localized perturbations. These findings suggest that focal interventions produce consistent yet region-specific patterns of global FC reorganization, providing an explanation for the seemingly paradoxical observations of hypo- and hyperconnectivity reported in the literature. This framework offers mechanistic insights into the systemic effects of localized neural modulation and holds potential for refining clinical diagnostics in focal brain disorders and advancing personalized neuromodulation strategies.