Tuesday, June 28, 2022

Navigating the Information Gradient

Olfaction is so primitive in its function, that it's an ideal model for all kinds of things,  including navigation, but even moreso, information processing. The olfactory system might be the most effective information processing system we know of, and it's something we've barely begun to investigate. 

Chemotaxis doesn't make headlines often, but it should, because it's ultimately an information-processing problem (and the last time I checked, we were living in the Information Age).

Image credit: A smellmap of Amsterdam by Kate McLean circa 2017 at sensorymaps.com

Information processing constrains how E. coli bacteria navigate chemical gradients
Jan 2022, phys.org

Information that E. coli bacteria gather from their environment limits their performance at chemotaxis, the process by which they guide their movements in response to chemical signals.

And it's funny that they decided to use chemotaxis to test this, about using information efficiently, so in other words, chemosensation is a good model for testing and understanding how information is processed, biomimetically, if you will.

And why do we care? Because chemotaxis and olfaction are the same, at a primitive level. Not much has changed between the way E. coli navigates its environment and the way we do it.

"We wanted to test a broad biological hypothesis: that organisms make the best use of the information they acquire to perform behaviors and other functions. To investigate this, we needed a behavior simple enough that we could quantify how much information it needed and chemotaxis by the bacterium E. coli is a perfect example of such a behavior."

We realized we could measure the amount of information a bacterium was able to gather (in bits per second), while also understanding how much information they would need to navigate at the speeds observed."

To achieve this, they first set out to calculate the theoretical performance limit, which is the maximum speed at which a bacterium could navigate up a chemical gradient, based on a fixed rate at which it acquires information about chemical signals.

Finding the response strategy that maximized gradient-climbing speed with a fixed information cost resulted in the performance limit.

"We found that while climbing shallow gradients E. coli get very little information from their environment, about 0.01 bits/s.

via Yale: H. H. Mattingly et al, Escherichia coli chemotaxis is information limited, Nature Physics (2021). DOI: 10.1038/s41567-021-01380-3

Understanding how bacteria seek out and move towards food
Feb 2022, phys.org

Chemotaxis is the process of attraction in the direction of a chemical gradient. The primary way that organisms control their motion and progressively move toward a target is by inhibiting tumbling when sensing that the chemical concentration is increasing along their current direction.

The research team used stochastic optimal control theory (instead of linear control theory) to find the best possible fully nonlinear sensing and control strategy of run-and-tumble motion (of E. coli) in environments with noisy chemical gradients.

And it looks like chemotaxis, which is the progenitor of olfaction. It is not a stretch to say that olfaction is a form of chemotaxis, and we move through a room to locate a source by using the pattern of its vaporized chemical essence in the air in the room. We calculate its distribution pattern (by stochastic optimal control theory, apparently^), predict the source, and move towards it, updating as we go. The only difference here is that we use legs, and a pretty complex limbic system, whereas E. coli just tumbles and tumbles in the chemovoid. 

via University of Tokyo Institute of Industrial Science: Kento Nakamura et al, Optimal sensing and control of run-and-tumble chemotaxis, Physical Review Research (2022). DOI: 10.1103/PhysRevResearch.4.013120

Odour-Spatial Map - Diogo Matias - Champalimaud Foundation - 2021 [link]

Neurons in the olfactory cortex link smells to places
Feb 2022, phys.org

Sometimes it's good to have someone else say things like this, for a change: 

The researchers focused on the primary olfactory cortex. "The olfactory system is unique among the senses," said the study's senior author, Zachary Mainen, a principal investigator at the Champalimaud Centre for the Unknown in Portugal. "Only olfaction has direct reciprocal connections to the hippocampal system, which is involved in memory and navigation."

It looks like neurons in the posterior piriform cortex (part of the primary olfactory cortex) are encoding place information just like hippocampal cells, and especially behaviourally significant spots. So it's real -- smells are not just smells, they are places and smells at the same time; we can't extricate them from each other, at least not for some brain cells.  

via Champalimaud Centre for the Unknown: Cindy Poo, Spatial maps in piriform cortex during olfactory navigation, Nature (2021). DOI: 10.1038/s41586-021-04242-3

Post Script:
How the brain navigates cities: We seem to be wired to calculate not the shortest path but the 'pointiest' one
Oct 2021, phys.org

When people navigate through a city, they use not shortest path, but instead, pedestrians appear to choose paths that seem to point most directly toward their destination, even if those routes end up being longer, and this is called vector-based navigation.

via  Massachusetts Institute of Technology: Paolo Santi, Vector-based pedestrian navigation in cities, Nature Computational Science (2021). DOI: 10.1038/s43588-021-00130-y

No comments:

Post a Comment