Naama Brenner (Technion) | Exploratory Learning in Biological Cells |


The capacity of cells and organisms to respond to challenging conditions is limited by a finite repertoire of repeatable responses. Beyond this capacity, novel and unforeseen challenges may elicit exploratory dynamics, improvisational in nature, potentially providing adaptation to a much broader array of conditions. Exploratory adaptation, its dynamics and convergence properties are not well understood.

Such phenomena are naturally described as learning processes: Learning entails self-modification of a system under closed-loop dynamics with its environment. In particular, the interactions between system elements are modified – like synapses during learning in the brain, that alter the connections between neurons. Inspired by classic concepts of Neuroscience, I will describe a theoretic framework for a primitive form of learning that takes place within the single cell. This “exploratory learning” is a random search guided by global feedback; we find this to be a feasible mechanism, but its convergence in high-dimensional gene expression space is non-universal and depends on network properties.

Successfully adapting network ensembles are heterogeneous and have outgoing hubs – the analog of “master regulators” in gene regulatory networks. The role of these hubs in guiding the search process is understood by a mapping to a simpler problem that can be analyzed by mean field methods and relates to a chaos-suppression phase transition. Results of this theory connect to several experimental observations in cellular systems. We thus establish a biologically plausible mechanism of adaptation by exploratory learning; this, in turn suggests that learning theory provides a useful theoretical framework to understand cellular dynamics.



When: March 2, 2022 02:00 PM (Israel Standard Time).

Where: Room 223, Multipurpose Bldg. & over Zoom