Many food intake events (= food uptake events) are too small for legacy “food intake measurement systems” or metabolic phenotyping systems to detect. Each of these feeding events corresponds to a neurological signal to feed, even if the actual amount is small. As such, they convey important behavioral information. I can easily imagine a gene knockout or a treatment that might affect micro-intake events while leaving macro-intake events unchanged – an important distinction to which most food intake measurement systems currently on the market are oblivious, but which Promethion can easily detect.
The list to the left, which is a small section of a food intake analysis spreadsheet, shows 7 food intake events from a C57BL/6J mouse. (Parenthetically, it’s interesting to see the reactions of people who are used to mouse food intake amounts expressed to the nearest 0.01 g or even 0.1 g, when they see this level of precision.) You can see that two of the events, highlighted, are below 10 mg.
This is where things get especially interesting. Legacy food or water intake measurement systems (which is to say, everything on the market except for Promethion) do not claim to detect food or water intake events less than about 10 to 20 mg. Promethion, on the other hand, was designed with fanatical attention to the highest possible resolution. In fact, intake events down to 2 mg can easily be detected. To achieve this level of resolution (about 1 part in 500.000) purely digital data transfer is essential. I blame my background in comparative physiology, concentrating on very hard-to-make measurements on small animals, for that emphasis on high resolution. Now, unexpectedly, it has opened a new window on the feeding behavior of laboratory mice. (Of course, this would also work for fluid or water intake.)
The first question that pops to mind is – are these tiny events actually real, or are they the result of random noise in the measurement equipment? Well, there is an easy way to test the random noise hypothesis. Each intake event is the result of comparing a stable mass before the intake event with a stable mass after the intake event. Each of those masses has a mean, a standard deviation and an N. Each is normally distributed. As a result, they can be compared using Student’s t statistic, which evaluates the probability of the two masses differing by chance.
And here you see the result. As you can see, the larger intake events have extremely high t values, corresponding to microscopically tiny probabilities (any probability below 0.001 is displayed as zero). But the micro-intake events also have a very respectable t values, demonstrating that there is no realistic probability that they are the result of random fluctuations in the measurement equipment. (This is also very obvious when looking at the raw data, which shows clear disturbances in the mass record during the micro-intake events; see below.)
If you’re interested, you can look at an image of a more complete section of the spreadsheet, automatically generated by the Promethion data analysis program, here.
But let’s do a belt-and-braces proof for you skeptics out there. To do so, let’s select the intake event denoted by the solid yellow bar at the bottom of the above image, and look at the raw data from which that data point was derived. This level of drill-down capability is, of course, unique to Promethion.
To the left, you can see a graph of the food hopper mass vs. time. This is raw data – no smoothing or any other processing was applied.
As the graph begins, the food hopper is untouched. Then the mouse starts to feed from the hopper, in the process exerting a downward force on the hopper that causes its measured mass to increase. (You might have noticed a column called UpF_g_min in the spreadsheet excerpt above; this is the integrated force that the mouse applied to the hopper during each feeding event, with the units g/min – which may be an indicator of motivational state, and is a measurement
The second question that pops to mind is, of course, who cares? The micro-intake events don’t contribute particularly significantly to total food intake! Why worry about them? Why not just ignore them? (Especially if you can’t measure them in the first place.)
I respectfully disagree. As I covered briefly in the introductory paragraph:
Each of these micro-intake events corresponds to a neurological signal to feed (even if the actual amount is small) and as such, it conveys important behavioral information. I can easily imagine a gene knockout or a treatment that might affect micro-intake events, perhaps by raising the satiation threshold, while leaving macro-intake events unchanged – an important distinction that traditional food intake measurement systems and metabolic phenotyping systems would miss.Promethion owes its ability to detect these intake events to Sable Systems’ many years of experience with ultrahigh resolution circuitry, and the use of load cells (as in lab balances) as mass transducers combined with the archiving of the entire raw data stream, which provides maximum flexibility of analysis.
I can think of many interesting research questions that arise from this. To take an easy example, are macro-intake events that follow multiple micro-intake events characterized by a slower intake rate? I see signs that that may be the case, but haven’t yet investigated this in detail. If you take the idea and run with it, good for you – I have more ideas than can ever be actually implemented.
I welcome your input on this, and any questions you may have. I can be contacted here.