Dr Avishek Das

Computations performed by living organisms are unique - they are error-prone yet low-energy and robust. How does such intelligence emerge, and how does it relate to biological function? Often, information processing and its biological use involve a combination of mechanical and chemical degrees of freedom, covering multiple timescales. My group will investigate the optimal designs of such mechano-chemical computational motifs using a combination of statistical physics and information theory. In this talk, I will highlight two stories in this direction: one about quantifying information, and the other about understanding its use in biological function.

The first is a new computational algorithm, called Transfer Entropy - Path Weight Sampling (TE-PWS), that makes it possible to exactly quantify the dynamical information transmission in any general stochastic system for the first time. The second story will be on stochastic navigators, i.e., navigators that sense and actuate with noise, such as chemotactic cells. Here, I will use theory and simulations to show that information feedback from the navigator to its environment is a crucial aspect of stochastic navigation, that there exists an optimal information feedback, and that information is necessary but not always sufficient. This study will thus explicitly derive the utility of information in stochastic navigators.