Astrocyte abacus
Another astonishing discovery in brain function.

Learning is controlled by an internal abacus.

Via Eurekalert, a study using larval zebrafish.

These two pictures tell the story.



The experimental setup is 'virtual reality' or in old-fashioned terms a flight simulator. The fish is held firmly in place in a small aquarium. A visual landscape is moved forward and backward in response to the swimming motion. In real life each wiggle moves the fish forward, and then the counterwiggle moves backward a bit. (Like a rowboat with incompletely lifted oars on the backstroke.) So the simulated landscape does the same thing.



As long as the landscape moves, the fish moves forward in bursts. Swim for a second, rest for a second. When the experimenters turn off the motion, the fish tries harder for about 30 seconds, then gives up entirely.

This is rational behavior for all animals in all** situations. Modern human dysculture aka "meritocracy" turns us irrational by persuading us that forward motion is ALWAYS necessary regardless of our observed gains relative to the status landscape. If you're not gaining, you need to try harder. We keep trying harder, wasting exponentially increasing quantities of energy and soul, until we violently burn out and die.

The experimenters were trying to determine how the fish decides when to quit. Is it just a buildup of charge or neurotransmitter on a comparator neuron?

What they found is ASTONISHING. The action is controlled by a specialized type of glial cell, not by neurons.

Here, we found that the fish analog of the mammalian astrocyte is a central computational element of a circuit implementing a behavioral-state change after integrating sensory information.

Specifically, radial astrocytes in a subregion of the brainstem in larval zebrafish temporally integrate noradrenergically encoded failures to accumulate evidence of futility before inducing a state of passivity (giving up). This behavioral pattern has a familiar combination of features: trying to achieve a goal, repeatedly failing despite trying harder, giving up temporarily, and then trying again. Fish swim more vigorously in open loop (i.e., ineffective swim attempts), become passive, then swim again.

First try harder, then give up for a while.

Using whole-brain imaging, we found that astrocytic calcium was elevated just before and during passive states. Activation and silencing experiments established that these glial cells are required to trigger the passive state. Neuronal imaging and manipulation showed that the NE system encodes an expectation-outcome mismatch signal, which activates Ca2+ signaling in radial astrocytes. Thus, behavioral failures are detected by NE-MO and integrated by glia, which, after accumulating sufficient evidence of motor futility, trigger a passive behavioral state via GABAergic neurons. Once passivity is triggered, its persistence may be due to sustained effects of the glial cells on neurons, or a lag in the reactivation of swim circuits.

In other words, the astrocytes serve as a kind of scorecard or abacus outside of the neurons. The neurons click up the astrocytes, and when the number of raised beads reaches a threshold the neurons tell the muscles to stop swimming.

= = = = =

** Later thought: It's obvious that frustration is universal, but why is it universal? Why do so many situations need this calculated give-up response? Specifically, why would zebrafish need to know when their attempt to swim didn't result in forward motion? Virtual reality experimental setups aren't exactly common in the real ocean. A headwind in the current would be the main reason for lack of progress. When you're moving against a headwind you do have to try harder, but I'd think the frustration response would include something like gliding down to the floor for stability until the current slows down, or turning around to swim with the current.

Labels: Grand Blueprint

  ¶ 12:27 AM