Want to hear about bugs, scents and secret codes?
My project for the next couple of lives is to understand how the brain works.
I am especially interested in understanding the neurophysiological basis of
awareness. Figuring that bacteria started to be understood before humans are, I am working on a system somewhat simpler than man: locusts.
The first question I'm asking about the brain is, what's the language that neurons talk to each other in?
The first part of that question has been answered: we know now that the letters of the alphabet of the language are mostly action potentials. However, we still don't know what the words are, or how many neurons need to speak at once for another one to listen, or how they coordinate with each other (as a chorus does via the director), etc.
Since modern biology has shown that men and worms share as much as Nietzsche had suspected, chances are the language is not very different from locust to man.
Some of the most interesting (to me, anyway) recent work about the spatio-temporal nature of the neural code used by
cells has come from the olfactory system of the locust at the Laurent lab. That's where I did my Ph.D. thesis.
Since to understand a machine it is useful to know what it does, Heather Dean (and previously also Yingzhong Tian and Libby Mosier, all undergrads in the lab) have been trying to find out what the locust does with its olfactory system. Thinking this might revolve around sex, we first did behavioral experiments to assess locust mating
preferences, and looking for a role for olfaction in that. That was not hugely successful, so we shifted to look at
their reaction to food odor. After initial preliminary reports by Ying and me that grass seemed to be attractive and
pentanol repulsive, Heather and I can now reliably obtain chemoattractive behavior toward grass and other odors and
repulsion from pentanol in a Y-olfactometer and an elevator maze (Backer, 2002, Chapter 3).
In addition, with the help of Yingzhong Tian, I have built Odogen (Backer,
2002, Chapter 6), an odor delivery system that is allowing me to look at the spatiotemporal coding of
odor concentration and odor mixtures in the antennal (olfactory) lobe of the
locust to better understand how the olfactory system solves the pattern recognition problem.
To look at the coordination between different neurons, I carried out an analysis of intertrial variability in the responses of
simultaneously-recorded neurons in the locust antennal lobe (Backer, Wehr and Laurent (1997) Soc. Neurosci. Abstr. 23). One of the conclusions was that neuronal effective connectivity seems to be changing dynamically during an odor response.
I have used a cost-based metric (Victor & Purpura, 1997) to show stimulus reconstruction (identification) in the olfactory system, proving highly reliable classification can be done by looking at a single projection neuron's spike trains (Backer, MacLeod, Wehr and Laurent (1998): Soc. Neurosci. Abstr. 24, 911). Kate MacLeod and I then used this technique to show that disruption of synchronization among antennal lobe projection neurons spares information in single PNs, but causes a loss of odor information in Beta-lobe neurons downstream (MacLeod, Backer and Laurent (1998): Nature 395, 693-698.) Here is a presentation
on it.
I have looked at how olfactory systems solve the probem of
pattern recognition, namely recognizing an odor despite variations in
the sensory input. To do this, I needed a systematic way to vary the
sensory input yet keep the same odor. I have done that by varying the odor
concentration. In looking at the responses of neurons in the antennal lobe
of the locust to different concentrations of the same odor, I
have made some interesting findings, including a surprising
learning effect. See Backer, 2002, Chapters 4, 5,
and 7-9 for details.
This page has been visited
times since June 1st, 2001.
Back