FED3 is a home-cage compatible device designed for high throughput training of mice
This is a battery powered FED device for training rodents in home cages. FED was developed to free researchers from expensive and rigid behavioral testing equipment, and allow them to do their experiments wherever they'd like - in dedicated boxed, in the home cage, or even inside other scientific equipment. Importantly, FED3 is open source and new sensors or code can be easily added to change its function.
The device runs on an Adafruit Adalogger M0 Feather board, which also includes a microSD card slot for recording behavioral data. The hardware include two "nosepoke" beambreak sensors, a beeper, a multi-color LED strip, a pellet dispenser, a BNC output jack for synchronizing with other equipment, and a screen for providing feedback to the user. See logs for info on the build, costs, etc. And as always, feedback is welcome!
Please join our Google group for troubleshooting and questions.
Funding and license:
This project was funded by the NIH Intramural Research Program (NIDDK). This project is released under the terms of the Creative Commons - Attribution - ShareAlike 3.0 license:
Newest code, includes programs with active/inactive pokes reversed, and also a self-stimulation program for use with optogenetic stim
Bug fix for FED3 getting stuck in sleep mode without a pellet in the well
The last release of STL files was in October of 2018, and I have made many tweaks since then to improve the function of FED. I have just uploaded a new set of STL files. These parts result in better dispensing, better aesthetics, and they solve some unexpected problems (like mice stealing SD cards!)
These include tweaks to all of the parts, which I'll quickly review here:
1. Wall Mounts. The front now includes corner insets to hold 8mm magnets for wall mounting. This allows FED3 to be mounted externally on a cage or wall, or be placed inside of it:
Here are 8 FEDs mounted on plastic boxes in my lab. Wall mounting keeps the FEDs cleaner but it requires dedicated space to run them.
2. Option for a metal pellet well. I included 2 variants of the front file, one (above) that is printed in one piece, as before, and another (in the STL pack) that has the "nose" removed. This is so the nose can be printed as a separate part in metal to stop mice from chewing on it (I've been doing this commercially through Shapeways at $15 each for the metal part). Here it is in brass (nose-piece STL is in the files area too):
3. SD card door. The housing now has a sliding door for covering the SD card and USB port (take that, curious mice!)
4. Larger diameter pellet disk. The diameter of the pellet disk was increased by 2mm to reduce the chance of pellets getting in between the disk and the wall of the pellet hopper, creating a jam. During assembly, it is important to "center" the disk so it rotates freely.
5. More reliable hopper. Four significant changes were made to the hopper itself. First, the pellet "foot" was decreased in size to just cover the dispensing hole. I found this reduced pellet "grinding". (Fun fact, this was like this in the FED1 design!)
Second, I removed the inner "ring" inside the pellet disk. This was designed to stop pellets from becoming jammed between the pellet disk and the wall of the pellet hopper, but it ended up creating a new place where pellets could become jammed if the hopper was not seated perfectly. With the wider diameter pellet disk this is no longer necessary. New (left) vs. old (right):
Third, I added "feet" to secure the orientation of the hopper correctly. When screwing the hopper on, the feet should align like this:
Fourth, I reduced the height of the hopper by ~1cm, which will allow FED to fit in even more caging (we had some caging that the old design was just barely too tall to fit in with the lid comfortably closed. In terms of capacity, the hopper previously fit ~2000 pellets, and now holds ~1500. This is an estimated drop from ~13 to ~10 days of dispensing, which I think is worth it for the decrease in height. FED3 needs to be checked more often than that anyway so it's sort of a moot point.
Old (left) vs new (right):
6. Tessellated lid text! This is purely aesthetic, but I redid the "FED3" text on top of the lid, using a tessellated design. As this lid is threaded part, the old "FED3" text would often end up slightly off center due to variance in the prints and it would really annoy me. This version can be rotated a few mms in any direction and it still looks nice :)
7. Tweaked back cover. Very minor tweaks to improve how it slides, and update the label to FED3 version 1.1!
If you made it this far here are photos of completed FED3 with the new parts:
Many of you may be using the FED3 in satellite mode, logging the data locally and extracting it for later analysis. Lex and I wanted to codify a platform where the outputs from the FED3 could be read in real-time and paired with other neural circuit manipulations like fiber photometry that use open source platforms like Bonsai. Briefly, Bonsai is an open source program with near limitless potential to read out and drive multiple pieces of hardware, software and provide you with behavioral or physiological data in real time. All for the low low price of $0. It uses a visual programming language to control different functions which means zero front end coding - the flow of commands appear as nodes that move left to right. Check out and install Bonsai here at https://bonsai-rx.org/
My goal was to use the FED3 as an operant feeding device reading out the sensors in real time to generate time stamps that I could then extract from fiber photometry data sets. I'm using the Neurophotometrics (NPM) System with 3 LEDs 415, 470, and 560 and a BlackFly Camera. Sage Aronson of NPM provided us with some starting bonsai files to control his system and I've adapted it and combined it with the FED3 bonsai layout. You can access the Bonsai layouts in the Files section above. Ultimately it does 3 things: 1) Talk to the FED3 and generate a timestamp file of the different sensor outputs 2) Talk to another camera to capture video of the chamber and 3) Talk to the NPM and generate fiber photometry signal data in csv format. A simple schematic of the hardware is below. Notably, you will need a separate Arduino to function as an I/O Box for the FED3.
We've recently been re-evaluating the utility of mounting FED externally to a cage or behavioral chamber, rather than placing FED3 in the cage itself. To faciliate this, we modified the front plate to include 4 magnet holes/mounts, and created a drill guide pattern for modifying the wall of a chamber to mount FED3 to it. We also used the "mini-hopper", which is available in the files area.
For mounting on a chamber or cage, it only requires drilling four 5/16" holes (for magnets) and two 2" holes (with a hole-saw) for the pokes and feeder. I was able to drill these and insert magnets in about 10 minutes. Video below, please contact me for modified FED3 front plate design and drill guide.
The Achilles heel of any pellet dispenser is pellet jamming. There are several reasons for a jam, but one of them is that the pellets inside the hopper align in a way that impedes the movement of the dispensing disc. To combat this issue, I've modified the stepper code to make the stepper "vibrate" while dispensing. This can help shake up the pellets in the hopper and reduce jams.
Please leave feedback in comments if you try this update!
Our lab is interested in how exposure to obesogenic diets might influence the willingness to work for palatable food, or to learn how to acquire such rewards. One way to test aspects of this is to perform an experiment in which a mouse learns to associate a specific task with the delivery of a reward. Since the amount of the reward and the work it takes the mouse to earn it are static in this experiment, it is called a "fixed ratio", or FR. We can then look at how quickly mice who are obese formed these associations compared to lean mice. Here are the weight curves showing how the "obese" group quickly gained weight on a high fat diet.
To test whether the obese and lean mice learned to acquire rewards (in this case a sugar pellet) differently, we used FED3 to launch a large-scale FR experiment between lean and obese mice. As you can see, the FED3 easily sits inside the home cages which are then housed in a rack in the vivarium.
(If you look closely you can see some of the cages with green tape on them labelled, "HFD," for high fat diet.)
We let the FED3s (running the Fixed Ratio program) run overnight, since that is when mice are most active. The next day, we pulled that data from the SD cards in each device, and used python to organize and plot our data.
As you can see, the lean mice learned to acquire more sugar pellets than the obese mice. This kind of straightforward experiment would have been very time consuming without FED3, which let us collect this data overnight!
I was asked by another researcher about mounting FED to a wall, instead of placing FED in a cage. It is very do-able but requires a modification to the hopper to make it a bit smaller so it doesn't stick out in the front. STLs for the modified parts are in the files area here. With the smaller hopper, FED can be mounted on the outside wall of a cage like this:
I also went ahead and printed a wall to fit in our operant boxes. This wall divides off part of the box so the mice can't get behind the FED. I wasn't sure how this would work but it came out pretty nice. The wall had to be printed in 2 parts because of its size. It was a fun modification, STLs for the wall parts are also in the files area here.
Our lab is completing an experiment looking at long-term effects of obesity on motivation to work for food. The question is, "are obese mice more or less willing to work for food than lean mice?" One way this can be tested is with a "progressive ratio" schedule, in which the first pellet requires one poke, the second two pokes, the third four pokes, etc. We can ask when the cost will be too high to continue. You can read more about this here.
To test this in our experiment we needed 20 FEDs. So we decided to build them! It was the largest FED build we've attempted and it went super well. Building 20 is different from building 1, and we learned some things that I'll adopt in the design to make manufacturing easier. But we got 20 FEDs built in ~2 days. Photos from the build below:
Before starting with building anything you should make sure your computer can communicate with the Adalogger M0 board and that you can flash the operant FED sketch. Start by installing the Arduino IDE.
Install relevant Arduino boards to run the Adalogger M0
The Adafruit M0 Adalogger board is not natively supported by the Arduino IDE. To enable the Arduino IDE to flash sketches to this board, follow instructions here. After completing these steps make sure you can flash the example sketch "Blink" to the Adalogger board before continuing. Open Blink in File>Examples>Basic>Blink. "Double-click" the button on the Adalogger to put it in bootloader mode and make sure Board is set to "Adafruit Feather M0" and port is set to the same before clicking upload (the right arrow at the top of the Arduino IDE).
You should see "Upload complete" in the bottom feedback window and the red LED on the board should blink once per second.
Congratulations, you have configured your Arduino IDE and uploaded a sketch to the Adalogger!
You can install them in multiple ways, but the simplest will be to download the prepared zip file (oFED libraries.zip, also in the files area) including these libraries and extracting it to your Arduino libraries directory (for most it is: /Documents/Arduino/libraries
When you're done they should all be in your libraries directory, see example:
If this doesn't work, or you'd prefer a different way go through instructions here for how to manually add libraries.