Jeff Gore '99

Research Interests

The Gore biophysics laboratory integrates experiment and theory to explore how interactions within a population or community can lead to rich emergent phenomena. Of particular focus is the use of laboratory microcosms to gain insight into the loss of stability and resulting chaotic fluctuations, alternative stable states, community assembly, and the spread of “cheater” strategies. Recently we have also become interested in the physics of learning in neural networks.

Loss of stability and the onset of chaotic fluctuations– Classic theoretical work dating back to Robert May has argued communities that multispecies communities will lose stability when there is an increase in either the number of species or the strength of interspecies interactions. Using hundreds of laboratory microcosms, we have experimentally demonstrated both of these key predictions and mapped out the theoretically-predicted dynamical phases of multi-species communities (Hu et al, Science (2022)). This built on previous work showing that pH-modifications of the environment can dictate many microbial interactions and that decreased buffering capacity relative to nutrient concentration leads to larger interactions and the onset of community fluctuations (Ratzke et al, Nature Ecol & Evo (2020)). These results demonstrate predictable emergent patterns of diversity and dynamics in ecological communities.

Alternative stable states and tipping points– Natural communities can often exist in distinct compositional states in the same environment, and these alternative stable states can have dramatic health and ecological consequences. Moreover, transitions between these states in deteriorating environmental conditions can occur suddenly, and recovery after such a transition can be exceedingly difficult. Theory predicts that changes in the fluctuations of population abundances might provide advance warning that such a “tipping point” is approaching. Our group has used laboratory yeast populations to experimentally measure these theoretically predicted early warning indicators based on critical slowing down (Dai et al, Science (2012)). In addition to changes in the temporal behavior, we have also found that characteristic spatial patterns emerge near a tipping point (Dai et al, Nature (2013)). On an entirely different scale, we have also demonstrated that cell memory loses resilience to environmental perturbations approaching such a critical transition (Axelrod et al, eLife (2015)). In recent work we found that even modest cooperative growth dynamics within a species can lead to multistability (Lopes et al, Nature Comm (2024)). This work highlights the ubiquitous nature of multistability and tipping points across scales.

Predicting community assembly and changes with the environment – A major question in ecology is determining the rules that determine which species will survive together in a community and which species will be driven extinct as a community “assembles”. We demonstrated that the pairwise outcomes of interspecies competition can be used to predict with reasonable accuracy whether species will be able to survive in trio competition (Friedman et al, Nature Ecol & Evo (2017)). We have also studied how community composition will shift in response to a changing environment, showing that the concept of a mortality burden provides predictive insight into how communities will shift with temperature (Abreu*, Dal Bello* et al, Science Advances (2022)). This body of work demonstrates that even complex ecological communities can behave in predictable ways.

Evolution of cooperation – The conditions required for the initiation and maintenance of cooperative behaviors is a classic problem in evolutionary biology. How can cooperators survive when they can be taken advantage of by “cheaters”? As a Pappalardo Postdoctoral Fellow here at MIT, Jeff demonstrated that a population of yeast, which cooperate by secreting the enzyme to break down the sugar sucrose, can be taken advantage of by mutant “cheater” yeast lacking the gene encoding the enzyme (Gore et al, Nature (2009)). However, in contrast to traditional models from game theory (eg the prisoner’s dilemma), he found that the normal cooperative yeast could survive the invasion by these cheater strains due to preferential access that the cooperators have to the fruits of their labor. This coexistence between cooperation and cheating appears to be a common pattern, for example also occurring between antibiotic resistant bacteria that are breaking down an antibiotic and susceptible bacteria that would normally be killed by the antibiotic (Yurtsev et al, Molecular Systems Biology (2013)). In recent work we found that the susceptible bacteria that are protected by resistant cells evolve in surprising ways, in particular by evolving a slow-death phenotype called tolerance (Pearl Mizrahi et al, PNAS (2022)). In several systems we therefore find that cooperation can be “taken advantage of” by cheater strategies, but in many contexts the cooperators can survive due to preferential access to the fruits of their labor.

Physics of learning in neural networks – Advances in machine learning over the last decade, and in large language models over the last few years, have demonstrated the emergence of a new form of “intelligence”. In contrast to neuroscience, in neural networks we know all of the interactions between “neurons”, yet we still do not understand how they operate. We are excited to understand the origin of neural scaling laws, the dynamics of training on the high-dimensional space of weights describing the interactions within the network, and the nature of reasoning within these models.

In all of this work we perform theoretically-motivated experiments and experimentally-motivated theory and modeling. This is a research paradigm that has been fabulously successful in the traditional areas of physics, and we believe that this approach can provide powerful insight into living systems.

Biographical Sketch

Jeff Gore is a Professor in the Physics Department at the Massachusetts Institute of Technology who studies the emergent dynamics of complex communities. His interdisciplinary work has been recognized by a Schmidt Science Polymath Award, Allen Distinguished Investigator Award, NIH New Innovator Award, Sloan Foundation Fellowship, NSF Career Award, Pew Scholar in the Biomedical Sciences Award, and an NIH Pathways to Independence Award (K99/R00). Before starting on the faculty he was a Pappalardo Postdoctoral Fellow in the Department studying the evolution of cooperation. He received his PhD from the Physics Department at the University of California Berkeley in 2005 with the support of a Hertz Fellowship using single-molecule techniques to probe twist and torque in DNA. Prof. Gore has also been active in the community, leading the Interdepartmental Biophysics activities. He has also served as co-Chair of the MIT Microbiology Program, the US representative of the IUPAP Commission on Biological Physics, Chair of the q-qio program committee, and Academic Editor at PLOS Biology. In addition to his research efforts, he is also passionate about teaching and mentoring, and received the Buechner Teaching Award and the UROP Faculty Mentor of the Year Award.