A radio controlled rat for demonstrating place cells
Some weeks ago I did a project for the IST Austria demonstrating place cells at the annual Open Campus science faire. The station was very successful, with more than 100 visitors in 5 hours.
Photo: Patrizia Gapp
The place cell is a specific cell type that neurobiologists study in the rat brain. Put simply, we have certain neurons that are active only when we are located at specific place in an environment. Some nevrobiologists, like the Csicsvari group at the IST, are interested in how the behavior of these cells relate to learning and memory.
Place field. Each dot represent a "firing" by the neuron. From knowingneurons.com
The idea was to demonstrate this concept by letting the visitors control a radio controlled rat running around in a maze not unlike the one the scientists use. With help of a motion tracking system we would then simulate that a "neuron in the rat brain" would "fire" when the rat was in a specific "place field."
 |
The rat and the maze. Photo: Ola Nordal
|
 |
| Not so easy to see, but the running path is displayed to the left. On the right a corresponding neuron would flicker in sync with the sound of the cell firing. Photo: Patrizia Gapp |
 |
| Schematic of the set up |
Getting technical
The set up is really very simple. The radio control unit was a 20€ RC car where I just flipped off the plastic hood. The rat was a 10€ cat toy I got off Amazon. I built the tracing and display system of the station with Max/MSP/Jitter, using a cheap standard webcam as tracking unit. The sound of a firing cell was diffused from a standard computer speaker. The total cost of the station, including cabling, a car that I broke, and two toy rats that were too ugly to be used, was less than 100€. Probably the "experiment table" built in the local workshop was just as expensive.
Photo: Ola Nordal
The Max patch
The core part of the Max-system was motion tracking. I previously had good experiences with the cv.jit package by Jean-Marc Pelletier, and decided to use cv.jit.centeroids as the central tracking object. |
| Main patch, displaying the video image after crop and brightness/contrast adjustment + detected blobs. The main problem was do calibrate the system so reflections on the maze did not get defined as blobs. |
The user would "look for" a cell on the table by driving the RC rat around, and when the rat would be detected in the assigned space, the patch would start playing back the sound file. The loudness output of the sound file would determine when a red dot would be drawn on the running path, and a when the "corresponding neuron" would flicker on the display image.
 |
| The mechanism for letting sound level threshold trigger "spikes" or "fierings" |
I defined five areas on the board that each would be a "cell" or "place field." The first "cell" was rather big and easy to find. When having localized a cell, the user could choose to look for a smaller cell. All together five "place fields" were defined, so the whole thing was a bit like a computer game. The kids loved that.
 |
| The five "Place fields" |
The Open Campus day
Sadly I don't have any photos from the actual Open Campus day (I was on "kid duty"), but the station was highly popular. A logging unit built into the system recorded more than 400 "experiments." I estimate that each user ran 3-5 experiments, meaning that at least 100 people used the station the 5 hours it was open. The presenter from the Csicsvari group (Charlotte) did not have any chance what so ever to get a break, and actually the station was still surrounded by people one hour after the Open Campus day officially was over and the other stations already were packed down.The whole project was a whole lot of fun to do, especially since the IST Staff, the Csicavari group and the visitors were so positive. I hope to put it up again for next years Open Campus, and maybe develop it further so it can travel to other locations not requiring my presence to set it up.