“Understanding the energetic costs of biological computation”
Biological information processing is energetically costly; our brain consumes more energy per gram than our muscles. But while thermodynamic tools roughly capture the energy requirements of the mechanical work done by muscles, we have no equivalent framework to understand the energy requirements and efficiency of the information processing done by our brains. Fundamental bounds applied without reference to physical details typically predict costs of order the thermal energy, KBT per bit, at least six orders of magnitude smaller than measured costs of transmission between neurons. Here I will argue that costs can still be understood as arising from the need to beat intrinsic thermal fluctuations, but that practical costs can be much larger than KBT due to geometric constraints and the physical properties of the communication medium.
I will focus on the energy cost of communication between molecular components. Living systems use physically distinct media for communication; electrical currents mediated by ion channels are carried through the cytoplasm and are sensed by distant voltage sensitive ion channels. And second messenger molecules are produced locally and diffuse to be sensed by distant molecular receptors. Each of these processes are inherently dissipative and inherently noisy, corrupted by thermal fluctuations intrinsic to the medium. I will show that sending signals that are reliable over these fluctuations implies costs which are often orders of magnitude larger than the thermal energy, plausibly accounting for the high energy cost of running ion channels and other signal transduction machinery.
Host: Karsten Heeger