Experiment #1
Pocket koji
Our very first foray into human-body incubation was intentionally raw and simple. The idea was to test the absolute minimum required to grow koji with personal body heat. We began by combining a small amount of steamed rice with koji spores in a standard ziploc bag. This makeshift “incubator” was then placed in Rasmus’s back pocket for a full 48 hours, while he went about his normal daily activities. The outcome was largely unsuccessful, yielding minimal to no visible koji growth. Given the highly uncontrolled and rudimentary nature of this initial setup, it was nearly impossible to pinpoint the exact reasons for its failure—whether it was temperature fluctuations, insufficient humidity, or even external contamination. This inconclusive, yet important, initial attempt clearly signaled the need to step up our game and implement a more refined experimental approach for our next test.
In all experiments, Jasmin rice was used. It was overnight soaked, and then steamed for 45 minutes. It was then cooled down to room temperature, after which it was combined with 0.02% by mass of koji spores, mixed, and incubated at relevant conditions.
Experiment #2
Body heat
For our second experiment, we upgraded to a 3D-printed, egg-shaped container. Inside, we placed a commercially available Smart Life air temperature and humidity sensor—this time without any koji—to specifically test our hypothesis: could human body heat alone provide the necessary temperature for koji’s growth? We kept the egg suspended under an armpit, in direct contact with the skin. The results were positive! Though the sensor only provided hourly readings, it consistently showed that we were able to maintain temperatures within koji’s ideal range of 25-35°C throughout the entire experiment.
Experiment #3
Sensor testing
Experiments 3 and 4 ran in parallel over a two-week period, informing each other as we simultaneously tested incubation methods on our bodies and identified standard component sensors for an engaging, safe learning experience. To answer what variables were crucial for koji’s well-being, we designed an experiment to track its growth with various sensors. We began by measuring koji’s basic needs: temperature, humidity, oxygen, and rice moisture.
Learning that overheating inside the rice clump made koji prone to contamination, we added a temperature probe to monitor internal temperatures. Since mold color is a key visual cue for potential contamination, we also implemented an RGB sensor. This resulted in a triplicate setup, allowing us to test different koji strains.
During the testing phase of the 72h triplicate experiments we also learnt that we needed to add a capacitor to the AHT22+ENS160 sensor, since it was unable to give reliable readings otherwise. We also had to replace our moisture sensors with a oxidation-resistant variant, since we saw that after 72h of continuous measurement the initial sensors were corroding.
One significant takeaway was the temperature probe’s ability to detect even a 1°C difference between the air and the rice interior. This insight means users can be prompted to mix the koji, directly increasing their engagement. However, the color sensor proved inadequate; it consistently produced a confusing mid-experiment peak despite the koji visibly growing greener over time. This unreliable outcome, coupled with the constantly changing ambient light conditions for a portable device, led us to seek an alternative solution. We ultimately settled on a transparent growth chamber, produced via vacuum forming with food-grade polystyrene.
We were excited with the resulting outcome, as it not only solved the sensor problem but also significantly boosts user engagement by allowing direct visual observation of koji’s growth. With these findings, we had identified all the necessary parts for our KojiGotchi device.
In total, four experiments were conducted, which are annotated as first_try, second_try, third_try and fourth_try. first_try and second_try were failed for various reasons. The two graphs presented here come from third_try (temperature) and fourth_try (color). The shared folder includes the python notebooks used to visualise and process the data. The temperature readings were factored to be 30°C at the flat sector in the beginning. The rest of the readings were adjusted accordingly
Overview of the basic sensors.
Air Temperature & Humidity: AHT22
CO2 and VOC: ENS160
Moisture: HD-38
Overview of additional sensors.
Temperature: DS18B20
Color: TCS34725
Electronic setup schematic
Experimental setup in the incubator
Oxidated moisture sensor had to be replaced with a different one
AHT22+ENS160 had to be supplemented with a 100nF capacitor because otherwise the consistency of the readings would fail. The moisture sensor had to be replaced as well (the double fork on the left)
Temperature measurements of the probe submerged in rice and suspended in the air throughout the duration of the growth of white koji on rice. Average and standard deviation shown for n=3
Color sensor readings for red, green and blue diodes of TCS34725 captured during the green koji on rice growth experiment. Average and standard deviation shown for n=2.
Experiment #4
Proof of Concept
After identifying the most suitable sensors and settling on the growth chamber, we conducted our final proof of concept experiment. This iteration incorporated the transparent growth chamber and the Smart Life temperature and humidity sensor from previous tests. We diligently tracked koji’s progress with periodic photographs, carrying the device as described in Experiment 2. The result was a resounding success! Although the growth took slightly longer than in a traditional incubator, the resulting koji was just as fragrant and tasty. Our hypothesis was fully confirmed: the sensor readings showed the temperature remained perfectly within koji’s ideal range for the entire 60-hour duration, demonstrating the human body’s reliable incubating power.
