Academia | Blog

UNIVERSITY WEBPAGE

CURRICULUM VITAE

BSc – Biological Resource Sciences – University of California, Berkeley 1978-1982
MSc – Entomological Sciences – University of California, Berkeley 1983-1985
Chateaubriand Fellow – INRA – Montfavet, France 1985-1986
PhD – Pure and Applied Biology – University of London 1986-1989
Research Fellow – Centre for Population Biology – Imperial College 1989-1991
Researcher and Research Director – CNRS – ENS, Université de Paris VI 1991-2000
Research Director – CNRS – Université de Montpellier 2000-

PROFESSIONAL AND ACADEMIC HONORS

1997 Silver Medal of the CNRS
1998 Founding Editor, Ecology Letters
1998 Honorary Fellow, University of Wisconsin, Madison
2003,2006 Fellow, National Center for Ecological Analysis and Synthesis, Santa Barbara
2009 Miller Professor, University of California, Berkeley
2012- External Faculty, Santa Fe Institute
2013-2014 Fellow and Focus Group Leader, Wissenschaftskolleg zu Berlin, Germany
2015-2019 Visiting Researcher, Institute for Advanced Study, Toulouse, France
2018 Elected to Academia Europaea
2019 Edward P. Bass Distinguished Environmental Scholar, Yale University
2020 The Darwin Lecture to the Linnean Society and the Royal Society of Medicine
2021 Elected Fellow American Association for the Advancement of Science

CURRENT SEMINARS

The future of academic publishing
Cancer evolution: from cells to species and back
Phage-bacteria community coevolution: from basic science to applications in phage therapy
A biologist’s perspective of process and pattern in innovation

ALUMNI NOW IN ACADEMIA

Minus van Baalen (PostDoc)
Olivier Plantard (PhD)
Sam Brown (PostDoc)
Corinne Vacher (PhD)
Nicolas Mouquet (PostDoc)
Alexandre Roulin (PostDoc)
Michael Brockhurst (PostDoc)
Timothée Poisot (PhD)
Gül Deniz Salali (M1 and M2)
Clara Torres-Barceló (PostDoc)
Andrei Akhmetzhanov (PostDoc)
Robert Noble (PostDoc)
James Gurney (PostDoc)

See Google Scholar for recent publications

RESEARCH SUMMARY

My group has been successful in addressing what we believe are interesting and important fundamental scientific questions, and interfacing these with more applied concerns, in particular therapies to combat cancer and bacterial pathogens. Over the past five years, we have championed systems-thinking to studying individual and species interactions, applied to disease. Our major advances are to have found a preponderant impact of environmental variation in ecological and evolutionary dynamics, and to apply environmental and evolutionary perspectives to system performance, including intelligence and human interactions with artificial intelligence.

COVID-19

I became involved in COVID-19 research at the beginning of the pandemic, when governments worldwide were facing unprecedented uncertainty about how to respond to a rapidly spreading virus. Although the basic epidemiology of infectious disease was well understood, translating this knowledge into actionable public policy posed a major challenge. Early debates often framed control strategies in binary terms—either “lock down” or “let the virus run”—without a clear understanding of how intervention strength, timing, and epidemic state interact. Against this backdrop, my objective was to use simple modeling to clarify how non-pharmaceutical interventions shape epidemic trajectories, and to provide conceptual guidance that could support real-time decision-making under severe constraints. This led a study focused on suppression, mitigation, and adaptive control of transmission during the COVID-19 epidemic.

Based on my experience modelling and experimenting on host-pathogen interactions dating back to my PhD thesis, I began by examining the distinction between suppression and mitigation, and why this matters for epidemic control (Hochberg 2020a). Suppression is commonly understood as strong intervention designed to reduce transmission well below the epidemic threshold, whereas mitigation aims to slow spread without necessarily reversing it. However, these terms were often used loosely in policy discussions. In this study, I showed that suppression and mitigation differ not only in intensity but also in their dynamical consequences. Using numerical simulations of a simple epidemic model, I demonstrated that early, strong suppression can rapidly reduce the number of infectious individuals to low levels, placing the system in a fundamentally different and more controllable state. Once infection numbers are low, more moderate mitigation measures can be effective at maintaining manageable dynamics. By contrast, beginning with weaker mitigation when cases are already growing allows infections to accumulate, making later control far more difficult even if stronger measures are eventually applied. The key result was that epidemic outcomes depend jointly on the reproduction number R and the absolute number of infectious individuals, highlighting why timing is just as important as policy strength.

Building on this insight, I next turned to the question of what realistic short-term goals for epidemic control should look like under social constraints (Hochberg 2020b). While driving R well below 1 is epidemiologically desirable, it may be unsustainable over long periods due to economic disruption, population fatigue, and compliance erosion. I argued that, once infections have been brought down through suppression or early control, countries should aim to maintain R close to—but slightly below—1 as a pragmatic short-term mitigation target. I showed that near-threshold control can stabilize epidemic trajectories, preventing explosive growth while avoiding the costs of maximal suppression. Importantly, the analysis emphasized that R ≈ 1 is not a benign or “safe” zone: small deviations above the threshold can lead to renewed growth, especially if surveillance is imperfect or public health responses lag. The contribution of this paper was to recast R not as a static metric, but as a control variable requiring continuous adjustment and careful monitoring.

Taken together, these two preprints established a coherent conceptual framework for managing COVID-19 outbreaks in the absence of pharmaceutical interventions. The first clarified why early suppression can dramatically alter the control landscape by reducing infection numbers, while the second articulated how mitigation should be tuned once that low-infection state is achieved. Rather than advocating a single fixed policy, this work highlighted the importance of monitoring, feedback, and pragmatics in epidemic management. It also underscored that epidemic control is inherently dynamic: policies must adapt to changing epidemic states, rather than being chosen once and held constant. These ideas naturally raised the next question: if maintaining transmission near the threshold is both necessary and challenging, can flexible policies be designed that less socially costly than more uniform restrictions?

This question motivated my collaboration Joshua Weitz (Georgia Tech) and his group, which resulted in a study on disease-dependent interaction policies (Li et al. 2021). The central idea of this work is that traditional mitigation strategies typically reduce contacts uniformly across the population, regardless of individuals’ disease states. While effective at lowering transmission, such ‘en masse’ policies impose large economic and social costs. In contrast, policies that modulate interactions based on infection status—such as isolating infectious individuals more strongly while allowing others greater freedom—have the potential to achieve better health and economic outcomes simultaneously. Using a modeling framework that combines epidemic dynamics with economic costs, we explored how interaction rates could be adjusted dynamically in response to epidemic conditions.

A key methodological contribution of the Li et al. study was the use of feedback control principles to design robust policies. While optimal control theory can identify idealized strategies, such solutions often rely on precise knowledge of epidemic parameters and states, which is unrealistic during an unfolding outbreak. To address this, together, we developed disease-dependent interaction rules that respond to observable epidemic indicators, such as case numbers, rather than assuming perfect information. These feedback policies were shown to approximate the performance of optimal solutions while being more robust to uncertainty and delays. The results demonstrated that adaptive, state-dependent interventions can substantially reduce epidemic burden while preserving more economic activity than uniform mitigation, offering a practical extension of the threshold-based control logic developed in the earlier COVID-19 studies.

Overall, my work on COVID-19 bridged epidemiological theory and policy-relevant insight during this rapidly evolving global crisis. Beginning with clarification of suppression and mitigation, progressing to realistic control targets near the epidemic threshold, and culminating in adaptive, disease-dependent interaction policies, these studies collectively show how simple models can yield actionable understanding. Rather than prescribing specific interventions, my research on COVID provides a conceptual toolkit for thinking about timing, intensity, and adaptability in epidemic control. This contributes to a broader understanding of how societies can manage future infectious disease outbreaks when neither unlimited suppression nor uncontrolled spread is an acceptable option.

Evolutionary medicine

My group’s work on evolutionary medicine has been motivated by the idea that many of the most difficult medical challenges—chronic infections, antimicrobial resistance, cancer, immune-mediated disease—are not static problems but evolutionary ones. Pathogens, tumor cells, and host responses change through time, often in direct response to interventions meant to control them. Yet medical practice has traditionally emphasized eradication: eliminate the pathogen, destroy the tumor, suppress the symptom. When ‘cures’ fail, it is often because the target has evolved. Our aim has been to reframe disease control as an evolutionary and ecological problem, and to develop conceptual tools that help anticipate, steer, or limit evolutionary responses rather than merely reacting to them after the fact (in the next sections, this is expanded in the areas of cancer evolution and aging).

My work on evolutionary medicine stems from study I conducted on the biological control of insect pests and pathogens in the 1980s and 1990s. I revisited this research in the 2000s, applying it to evolutionary medicine starting with my interest on phage-bacteria interactions, phage therapy, and host-cancer interactions, and cancer therapy. Importantly, I was invited in 2015 to contribute a chapter to a book called Unsolved Problems in Ecology, where I began to develop a general framework for thinking about disease treatment strategies in evolutionary terms. This became a preprint and then a chapter entitled Six Wedges to Curing Disease (Hochberg 2020c). I argued that curing disease is rarely achieved through a single mechanism, and that most successful interventions implicitly rely on combinations of distinct processes, or ‘wedges’, that together reduce disease burden below a critical threshold. These wedges include suppressing pathogen or tumor populations, limiting their ability to transmit or spread, reducing damage to the host, preventing the emergence of resistance, and altering the ecological context in which disease occurs. A central contribution of this work was to generalize treatment beyond eradication, showing how durable control can emerge from partial, complementary effects that jointly constrain disease dynamics. By placing evolution explicitly at the center of this framework, this chapter highlighted why single-target, high-intensity interventions often fail.

Building on this conceptual foundation, I established a collaboration with Dr. Paul Turner during my 3-month stay at Yale University in 2019. Our work on phage steering addresses a central problem in infectious disease control: antibiotics impose strong selection for resistance, and once resistance evolves, therapeutic options rapidly narrow. Traditional strategies treat resistance as an unavoidable by-product of treatment, to be managed reactively by drug rotation or escalation. From an evolutionary perspective, however, resistance is a predictable outcome of selection acting on microbial populations rich in genetic variation. The question motivating these studies was therefore not how to eliminate evolution, but whether it could be directed. Specifically, we asked whether bacteriophages—viruses that infect bacteria—could be used not only to kill pathogens, but to shape the evolutionary paths available to them, biasing adaptation away from antibiotic resistance and toward more treatable states.

In Phage steering of antibiotic-resistance evolution in the bacterial pathogen, Pseudomonas aeruginosa (Gurney et al. 2020a), I directed a study testing this idea experimentally. Building on earlier observations that some phages exploit bacterial structures involved in drug resistance, we examined how exposure to a specific lytic phage altered the evolutionary response of Pseudomonas aeruginosa to antibiotics. We showed that phage selection can force bacteria into an evolutionary trade-off: resistance to the phage arose through modifications that simultaneously increased sensitivity to antibiotics. Crucially, this effect was observed both in vitro and in vivo. The study demonstrated that phages can be used deliberately to “steer” bacterial evolution, transforming resistance from an obstacle into a lever for control. Rather than merely suppressing bacterial density, phage treatment reshaped selection, constraining which adaptive solutions were accessible.

These experimental results were placed in a broader conceptual context in Steering phages to combat bacterial pathogens (Gurney et al. 2020b). This review that I led with a postdoctoral scientist in my group, Dr. James Gurney, articulated phage steering as a general strategy grounded in evolutionary and ecological principles, emphasizing that therapeutic success depends on understanding how selection pressures interact across time. We argued that phages should be chosen not only for their immediate killing efficacy, but for the evolutionary responses they elicit—particularly whether host resistance mechanisms impose collateral costs. The paper also outlined risks, such as evolutionary rescue and ecological complexity in real infections, while emphasizing that these challenges are tractable when evolution is treated as an integral part of therapy design. Together, these two Gurney et al. studies reframed antimicrobial treatment as an information-guided evolutionary process: by shaping the selective landscape, clinicians can influence how pathogens adapt, turning evolution from an adversary into a tool for durable disease control.

Together with collaborators at Yale University, I led a perspective article on how host-associated microbial ecosystems and tumor microenvironments shape both disease progression and therapeutic outcomes (Burmeister et al. 2021). This work addressed a key limitation of traditional antimicrobial strategies: they typically focus on the pathogen in isolation, ignoring the broader ecological community and ecosystem in which it is embedded. We reviewed evidence showing that host microbiomes influence pathogen growth, virulence, and evolutionary trajectories, and that treatments can have unintended consequences by disrupting these interactions. Importantly, the paper argued that evolution and ecology are inseparable in host–microbe and host-tumor systems: ecological changes alter selection pressures, and evolutionary responses feedback on ecological structure. Our study articulated an integrated framework in which disease management explicitly considers host ecosystem interactions as leverage points for controlling pathogen evolution.

As these ideas matured, it became clear to me that evolutionary medicine faces several structural challenges that go beyond any single disease system. In Addressing challenges in evolutionary medicine: three priorities (Hochberg 2023a), I argue that the field’s near-term progress depends on confronting three central challenges and pursuing three corresponding ways forward. The challenges are, first, determining how far evolutionary medicine can go in “nudging” life systems and/or their environments toward healthier states in an Anthropocene context, given that physiological and behavioral traits are interwoven, often individual-specific, and may entail underappreciated costs when modified; second, responsibly applying biomimetics, i.e., drawing on macroevolutionary innovations across the tree of life for human health while recognizing that traits are embedded in coevolved interactive systems and that transfers may be logistically difficult or carry downstream negative effects; and third, increasing engagement, because the uptake of evolution-instructed approaches is slowed by medical conservatism, clinical-trial barriers, industry hesitancy, and limited evolutionary training among clinicians and policymakers. The article then outlines three ways forward: expanding systems research that brings complex-systems thinking into both fundamental work and applications while identifying robust ‘system levers’ amid nonlinearity and uncertainty; strengthening involvement of clinicians and policymakers through sustained venues, training, and collaboration that make evolutionary reasoning usable in practice; and developing compendia that systematically document when evolutionary approaches succeed or fail across interventions and contexts—reducing reporting bias, enabling like-with-like comparison, and clarifying when evolutionary significance translates into medical significance.

Finally, stemming from an extended visit to the Freie University in Berlin, Germany in 2022, together with Dr. Jens Rolff, I co-wrote a novel perspective on fighting microbial pathogens in the form of forecasting antimicrobial resistance evolution (Rolff et al. 2024). Antimicrobial resistance is one of the clearest examples of evolution undermining medical control, yet most responses remain reactive, detecting resistance only after it has become widespread. In our review that included the input from several specialists, we assessed the current state of resistance forecasting, drawing on evolutionary theory, pharmacology, and clinical data. We showed that while short-term predictions are increasingly feasible, major challenges remain in linking drug exposure, pathogen population dynamics, and evolutionary responses across complex environments. The key contribution of this paper is to formalize forecasting as a necessary component of resistance management, arguing that interventions should be evaluated not only by their immediate efficacy, but by their predicted evolutionary consequences under realistic use patterns.

Taken together, these above studies form a coherent approach to reshape how disease is understood and managed. From the “six wedges” framework that redefines what curing means, through the integration of host ecology and microbial evolution, to strategic priorities and concrete forecasting challenges, this body of work emphasizes that disease control is fundamentally about managing dynamic systems. Rather than seeking permanent solutions through ‘hitting hard’, it advocates for informed, adaptive strategies that acknowledge evolutionary constraints. In doing so, my research contributes to a growing shift in medicine—from fighting evolution to working with it—so as to achieve more durable and effective health outcomes.

Cancer ecology and evolution

My group began working on questions relating to cancer evolution in 2009 after I stepped down as Editor in Chief of Ecology Letters. Entering a new discipline meant consecrating considerable time to integrating the published literature and it was during my sabbatical at the Wissenschaftskolleg zu Berlin in 2013-2014 that my cancer projects really matured and began producing interesting findings.

Cancer is fundamentally an evolutionary process occurring within the body, yet clinical oncology has long relied on coarse descriptors—tumor size, grade, and stage—that only indirectly reflect the underlying evolutionary dynamics. As genomic technologies have matured, it has become clear that tumors are not homogeneous entities but complex, evolving populations composed of interacting subclones. However, translating this insight into predictive or actionable understanding has proven difficult. My research program has sought to identify which features of somatic evolution matter most for clinical outcomes, how those features arise mechanistically, and how they interact across biological scales—from cellular behavior and signaling to organismal life history. This effort has led to a series of theoretical and empirical studies that aid in our understanding of cancer as an evolving ecosystem.

We began by addressing a simple question: when, why, and how does tumor clonal diversity predict patient survival? Although clonal diversity is often assumed to be a poor survival prognosis because of evolvability and treatment resistance, empirical associations between diversity and survival have been inconsistent across cancer types. Co-led with my postdoc Dr. Robert Noble, we applied evolutionary theory to mathematical modeling and empirical evidence to show that clonal diversity is not necessarily intrinsically prognostic (Noble et al. 2020a). Instead, its predictive value depends on when diversity is measured, why it has arisen, and how it relates to ecological interactions within the tumor. We demonstrated that high diversity early in tumor development can signal rapid mutation accumulation and future adaptability, whereas high diversity late in progression may reflect ‘competitive release’ or slowing growth. Our study’s core achievement was to move the field away from static interpretations of diversity and toward a more dynamic, process-based understanding that links diversity to the tempo and mode of somatic evolution.

This emphasis on dynamics naturally led us to investigate how somatic mutant populations change through time in response to perturbations, particularly in clinically tractable systems. In a collaboration focused on myeloproliferative neoplasms, I proposed a study to researchers who were part of a research consortium that I founded in 2011, where we examined how interferon-α (IFNα) therapy could reshape the population dynamics of mutated hematopoietic stem and progenitor cells (Mosca et al. 2021). These diseases are driven by somatic mutations that confer proliferative advantages at the stem cell level, yet patients often show heterogeneous and delayed responses to treatment. Using longitudinal patient data combined with mathematical inference, our team reconstructed the hidden dynamics of mutant and wild-type stem cell populations under therapy. We found that IFNα does not simply eliminate mutant clones; instead, it alters their growth and differentiation rates, sometimes inducing transient expansions before longer-term declines. This study provided one of the clearest empirical demonstrations that therapeutic effects on somatic evolution can be indirect, delayed, and state-dependent, reinforcing the need for dynamic rather than static biomarkers of treatment success.

While these studies focused on diversity and population trajectories, they left open a crucial question: how do tumor subclones actually interact with one another? Cancer evolution is often modeled as competition for space or resources, but tumors are also rich signaling environments in which cells modify their neighbors’ behavior. To explore this, I initiated and co-directed an investigation into paracrine interactions between tumor subclones: we asked whether such interactions could generate ‘parasite-like’ dynamics within tumors (Noble et al. 2021). Using theoretical models grounded in ecological interaction theory, we showed that subclones can evolve strategies that benefit from the secretions of others while imposing net costs on the tumor population as a whole. These interactions resemble parasitism, not because one clone is genetically ‘foreign’, but because it exploits shared signaling environments. The key insight of our study is that intra-tumor heterogeneity is not merely a reservoir of future adaptations – it can actively structure current tumor behavior through evolved clonal interactions.

The recognition that tumors and their microenvironments function as ecosystems, structured by interactions rather than just mutation accumulation, opened the door to broader ecological thinking. I had be working on this concept for several years, when I was contacted by my long-term collaborator, Dr. Pablo Marquet (University of Santiago, Chile) to pursue an interesting question. We applied an ecosystem perspective in a study of metastasis as a stoichiometric and network-embedded process (Castillo et al. 2023). Rather than treating metastasis solely as a property of particularly aggressive cells, we asked how metastatic cells exploit resource imbalances and interaction networks across tissues. Drawing on ecological stoichiometry and network theory, we showed that metastatic success depends on how well cancer cells match the elemental and metabolic constraints of target environments. Metastatic cells are not universally superior competitors; they thrive in particular niches created by organism-wide resource flows. Our work reframed metastasis as an emergent property of cancer–host interactions, linking cellular traits to systemic constraints, and reinforcing the view of cancer progression as a multi-scale ecological phenomenon.

Across these studies, a common theme emerged: somatic evolution cannot be understood in isolation from organismal biology. Mutation rates, selection pressures, and ecological interactions within tissues are themselves shaped by host life history, physiology, and disease processes. This realization motivated a more integrative effort to link somatic mutation dynamics with broader concepts of aging, disease burden, and organismal fitness. In Disease as a Mediator of Somatic Mutation–Fitspan Coevolution, I developed a theoretical framework in which disease both results from and feeds back on somatic mutation processes, influencing the length and quality of the functional lifespan—or ‘fitspan’—of the organism (Hochberg 2025b). The model formalizes how selection can act on somatic mutation rates indirectly through their effects on disease dynamics and survival, providing a unifying perspective on cancer, aging, and age-related disease. The model predictions also are consistent with empirical data across mammal species.

Crucially, this study does not treat cancer as an exceptional pathology, but as one manifestation of a general evolutionary trade-off between somatic maintenance and functional performance. By integrating empirical tests with model analysis, my work shows how disease can mediate feedbacks between somatic mutation accumulation and organismal fitness, potentially explaining why mutation rates are neither minimized nor maximized across species. This perspective connects directly back to earlier findings on clonal diversity and tumor dynamics: diversity, interaction structure, and treatment response are all downstream consequences of how somatic mutation processes are regulated over the life course. In this sense, cancer becomes a window into broader principles governing somatic evolution and life history evolution, rather than a special case requiring ad hoc explanations.

Taken together, this body of research advances a coherent view of cancer as an evolving, interacting system embedded within a living host. Beginning with clarification of when clonal diversity is informative, progressing through dynamic inference of somatic populations under therapy, uncovering parasite-like interactions among tumor subclones, and extending to ecosystem-level and life-history perspectives, these studies collectively move the field beyond static description. They show that prediction and control of cancer progression require attention to timing, interaction structure, and cross-scale feedbacks. Rather than seeking universal rules, my research emphasizes understanding the processes that generate heterogeneity and shape evolutionary trajectories—an approach that is essential if evolutionary thinking is to meaningfully inform oncology and medicine.

Aging and disease

Much of my research on cancer involved the question of aging, since cancer is ultimately a disease that takes time to develop and is more likely to develop as homeostatic mechanisms deteriorate, that is, age. I began a collaboration with Dr. Daniel Promislow (Tufts University) in 2018 that blossomed into a series of three working groups at the Santa Fe Institute and a second, (still on-going) series of working groups at the Santa Fe Institute and the Complexity Science Hub in Vienna. These two series of workshops ultimately stemmed from a previous series that I organized with Dr. Jeremy van Cleve (University of Kentucky) starting in 2016 on social dilemmas!

Aging and why organisms (generally, but not always) go into decline and die, has attracted concerted attention as an evolutionary problem, at least since the work of Medawar, Williams and Hamilton in the 1950s and 1960s. Evolutionary biologists have emphasized trade-offs, demography, and declining force of selection with age; biomedical scientists have focused on molecular damage, dysregulation, and disease; demographers and ecologists have examined survival curves and population heterogeneity. As data have exploded—from longitudinal cohorts and high-resolution omics to experimental manipulations in short-lived species—these perspectives have become harder to reconcile within a single explanatory framework. The two publications that I have contributed to so far, address this fragmentation. They seek to update evolutionary thinking on aging by introducing integrative concepts that connect mechanisms, life histories, and outcomes across scales, while remaining compatible with growing empirical evidence.

In Resilience integrates concepts in aging research (Promislow et al. 2022), Dr. Daniel Promislow (Tufts University), Dr. Rosiland Anderson (University of Wisconsin), and I co-led a study, where we argued that many apparently disparate phenomena in aging can be unified by focusing on ‘resilience’: the capacity of an organism to resist, absorb, and recover from perturbations. Rather than treating aging solely as the accumulation of damage or the decline of specific functions, this paper reframed aging as a progressive loss of resilience across physiological systems. Drawing on examples from ecology, medicine, and complex systems theory, we showed how resilience naturally links stress responses, homeostatic regulation, disease vulnerability, and mortality risk. Importantly, resilience is dynamic and measurable: it can be inferred from recovery rates following perturbations, variability in physiological signals, or responsiveness to stressors. The key achievement of our study was to provide a common language that allows evolutionary theory, clinical observations, and systems biology to speak to one another, without privileging any single mechanism as the principal cause of aging.

These ideas were extended and placed in a broader evolutionary context in How and why does aging occur? Updating evolutionary theory to meet a new era of data (Metcalf et al. 2025). The paper is the fruits of working groups I co-led with Dr. Jessica Metcalf (Princeton) and Dr. Daniel Promislow (Tufts University) at the Santa Fe Institute. Our paper addressed a growing tension in the field: classical evolutionary theories of aging remain conceptually powerful, but many were developed when data were sparse and largely cross-sectional. Today’s datasets capture within-individual change, social and environmental modulation, and molecular processes at unprecedented resolution. Our paper reviewed how core evolutionary principles—declining selection with age, trade-offs, mutation accumulation, and antagonistic pleiotropy—remain essential, but must be reformulated to accommodate this richness. A central contribution of our work is to emphasize that aging is not a single process but an emergent outcome of interacting physiological, ecological, and evolutionary dynamics unfolding over the life course.

Crucially, the 2025 paper argues that concepts such as resilience provide a bridge between evolutionary theory and data. By focusing on how selection shapes the regulation and recovery of systems, rather than only their long-term deterioration, evolutionary models can better align with observed patterns of variability, reversibility, and context dependence in aging. Our paper also highlighted the importance of integrating disease processes, social environments, and somatic evolution into evolutionary accounts of aging—an approach consistent with the resilience framework’s emphasis on perturbation and response. Together, these two publications move aging research away from single-cause explanations and toward a synthesis in which aging is understood as the progressive erosion of adaptive capacity shaped by evolutionary history. In doing so, they provide a conceptual foundation for interpreting new data and for designing interventions that aim not merely to extend lifespan, but to preserve health-span.

Information: from microbes to human-AI interactions

Throughout my career, I have been fascinated by the possibility that there is something fundamental beyond molecular (generations) and cultural (within and between generations) evolution and adaptation. It struck me during a webinar workshop that I co-organized in 2022 “Constructing and deconstructing collectives: Signals to space and society” that the key was information sensing, processing, and behavior—that is, ‘intelligence’. Across biology and technology, individual or collective adaptive success depends on how information is gathered, filtered, shared, and acted upon. Microbes coordinate behavior using chemical signals; humans rely on social learning, institutions, and technologies; and modern human societies are now entering an era in which artificial intelligence systems participate directly in information processing and decision-making. Despite some obvious differences, these systems face a common problem: individuals possess only partial, often noisy information, yet are selected to act in ways that are adaptive. My research is beginning to address this missing piece in the continuum of selection over molecular, cultural, and lifetime scales. I am exploring how information can be aggregated across individuals, how such aggregation generates adaptive intelligence, and how new forms of information-sharing—especially between humans and AI—may give rise to novel evolutionary units.

I was contacted by Dr. Stefany Moreno-Gamez (MIT) in 2020 to contribute my theoretical expertise on a study examining quorum sensing as a paradigmatic example of collective information processing in microbes (Moreno-Gámez et al. 2023). Quorum sensing allows bacteria to modulate gene expression in response to the density of conspecifics, using diffusible signaling molecules as a proxy for local population size. Traditionally, quorum sensing has been interpreted as a coordination mechanism enabling synchronized behaviors such as biofilm formation or virulence. In our study, we reframed quorum sensing as a way to harness the ‘wisdom of the crowds’: using evolutionary modeling, we showed that individual cells can reduce uncertainty about their environment by integrating their own noisy signal production with the signals emitted by others. This collective averaging improves decision accuracy, even when individual measurements are unreliable. The key achievement of our work was to demonstrate that quorum sensing is not merely a switch triggered at high density, but an adaptive information-aggregation strategy shaped by natural selection to exploit distributed knowledge within microbial populations.

This microbial perspective provided a foundation for a broader theoretical synthesis I was developing at the time. In the paper An Information Framework of Intelligence (Hochberg 2025a), intelligence is defined not by cognition, consciousness, or problem-solving prowess per se, but by the capacity of a system to acquire information, transform it into internal states, and deploy it adaptively toward goals. Crucially, this definition applies across scales and substrates: from bacteria responding to chemical gradients, to humans navigating social environments, to artificial systems. My framework formalizes intelligence as an emergent property of information flows and feedbacks, rather than a property tied to particular system architectures. By abstracting away from mechanism and focusing on information dynamics, my paper provides a common language for comparing intelligence in microbes, organisms, collectives, and machines.

These ideas culminate in an explicitly evolutionary treatment of human–AI systems in Could humans and AI become a new evolutionary individual? by Rainey and Hochberg (2025). Dr. Paul Rainey (Max Planck Institute, Ploen, Germany) read my article on the information framework of intelligence and contacted me to collaborate on an opinion piece for a widely-read journal. Our central question was whether tightly coupled human–AI collectives could undergo a major evolutionary transition, analogous to the emergence of multicellularity or eusociality. Drawing on evolutionary theory, we argue that such a transition would require stable information integration, aligned incentives, and mechanisms that bind human and AI components into a shared adaptive unit. Importantly, the paper emphasizes that information exchange is the critical bottleneck: without reliable, reciprocal information flows, hybrid systems cannot function as coherent individuals. The contribution of this work lies in reframing human–AI interaction not merely as tool use, but as a potential evolutionary process driven by shared information processing and selection on collective performance. Attesting to the interest in our viewpoint, our OpEd has been consulted over 30,000 times in the past 6 months.

Taken together, these studies trace a continuous thread from microbial populations to human–AI hybrids, unified by the adaptive role of information. Quorum sensing shows how even simple organisms exploit collective signals to improve decision-making under uncertainty. The information framework of intelligence generalizes this principle, revealing intelligence as a property of systems that manage information effectively across scales. Finally, the evolutionary analysis of human–AI collectives extends these ideas into the future, suggesting that new forms of individuality may emerge wherever information is sufficiently integrated and adaptively deployed. Across microbes, humans, and machines, the message is the same: intelligence is not confined to minds or brains, but arises wherever information is pooled, processed, and used to shape adaptive behavior.

​