What happens when you are trying to modify your behavior with information that comes from measurements that can be difficult that feed into models that are fundamentally inaccurate even though the you may not suspect the models are flawed? How do you understand this well enough to get a useful signal in a very noisy environment?
For many of us the period between Thanksgiving and the New Year involves way too much food. I’m trying to maintain my weight after a serious ten month diet and this means measuring and tracking what I eat as well as my exercise.
It should be simple - after all, energy is always conserved. Food is just a way to store energy. If you eat as much as you require to stay alive, your weight should be stable. Eat a bit less and/or do a bit more physical work and you should lose weight. Take in more than you need and your body tries to hold onto the extra energy by turning it into a energy stored in you - body fat.
But to do this simple calculation requires you have to know how much of the energy in food is accessible to your body, how much energy your body uses to run itself and how much energy it takes for you to do any kind of physical work. All of these are difficult to know to a good degree of precision. We have published guidelines, but few of us consider the error bars on the underlying measurements and calculations and some of the pieces are still being discovered by basic research.
We tend to assume food nutrition labels are accurate, but it turns out they are based on a technique that is about one hundred years old. That system, the Atwater System, has known flaws, but nothing has been found to be easy to use and more accurate.
Wilber Olin Atwater was a professor at Wesleyan University and a central figure in understanding modern nutrition. By the mid 1800s people had worked out that proteins, fats and carbohydrates were major components in human nutrition and estimates were made of how much energy was available from each through metabolism.
Atwater spurred the creation of an American nutrition research enterprise - one of the first examples of government funded big science outside of the military - and calculated how much metabolisable energy was available in proteins, fats and carbohydrates accounting for the efficiencies of digestion. It became possible to measure the chemical energy in food components and predict how much we could use. You could simply measure the amount of protein, fat and carbohydrates in a food and determine how much energy the human body would extract.1
The current standard used in the US and many other countries takes into account a few other nutrition sources and food labels are based on the following conversions:
It turns out this is only a general approximation and to first order you can get away with just looking at the fats, proteins, carbohydrates and alcohol figures if you are creating your own recipes.
But your efficiency at absorbing food varies. New work that looks at how the food was prepared suggests that food that is mechanically made into smaller bits is more efficiency absorbed and food that is cooked is absorbed even more metabolised.2 The largest effect is cooking. Muscle proteins, sugars in starches and other energy laden molecules are physically opened up allowing digestive enzymes to attack their amino acid chains more effectively. Food that is cooked gets more of its energy into your system than food that isn’t. The science is still developing, but some researchers speculate cooked foods provide at least 10% more energy than raw foods and the number may be higher than 25% - I wouldn't be surprised given some of the preliminary results.
The effect of cooking is not accounted for by nutrition labels. So if you want to lose a bit of weight you might try increasing the raw food component of your diet.
The energy balance for your body has four general components: energy intake from eating, energy expended keeping your body running, energy spend doing physical work, and energy spent in thermogenesis (shivering for example). Physical work comprises about 20% of the average person's energy budget, thermogenesis is usually under 10%. Most of the energy we use goes into keeping our bodies functioning - our Basal Metabolic Rate (BMR).3
Once you have an idea what your BMR is you can add up all of the physical activity during a day and have a fair idea what your intake should be - assuming you have a good idea of how much energy your body is getting from food. This gets tough, so most people who worry about this talk in terms of basic activity levels that range from 1.2 for sedentary office workers to around 2.0 for people with physically active lives.4
My BMR approximation is 1,583 kcal a day by the basic calculation. Colleen’s is 1,724, but her real figure is higher as she is very athletic.5
If you exercise regularly you can calculate the energy spent during exercise and add that to your BMR for a rough approximation - a lower limit as you move around doing other things during the day and there is some thermogenesis. Unfortunately energy spent during exercise is only an approximation as our efficiencies vary person to person for various exercises. That said you can probably get 80 to 90% accuracy using tools like this one.6
If you have made it this far the act of journaling your calories and exercise appear to be a futile effort as there is a huge amount of underlying complexity. But you can get around that.
Keep track of your weight too. If you plot the results you can get an idea of what your measurements for activity and eating mean assuming the types of food you eat and the activities you do don’t vary dramatically. The numbers may not match what you would measure in a lab, but getting the bottom up numbers to match the top down results is extremely difficult even in that setting.
Absolute numbers are very difficult to measure in this case, but it is possible to come by a stable relative measure, and that is all most of us need for this purpose.
My goal was to lose about 5 pounds a month while dieting. Using the rough figure that a pound of fat (what I wanted to lose) is about 3,500 Calories, I aimed at an average daily deficit of about 625 Calories through a combination of exercise and eating less food. It worked and I was able to maintain a fairly constant rate until near the end when I wanted to gracefully transition to an exercise/eating routine that might be sustainable.7
Even though we have flawed models for how much energy we absorb from our food and how efficient our bodies are, the models are sufficient - with a bit of analysis - to effectively use as a tool for losing and maintaing weight. It is easy to imagine local refinements - I’ve created a tool that helped me understand my own information flow and ultimately how that translates into weight - and it is likely that activity monitors like accelerometers can help, but in the end simple willpower rules. I suspect many monitoring and computational aids will ultimately fail unless they can impact our basic urges. To control something you need to combine motivation, the ability to make a change and some triggering event. I’ve worked that out for myself. Building a general tool seems difficult.
Perhaps what we really need to do is find the levels of eating and activity that evolution assumes. Up until the Industrial Age people appear to have been self regulating in weight. Now foods tend to be more plentiful, have greater energy densities and perhaps are even more palatable. At the same time many of us are less active. The balance is thrown and many of us gain weight easily. Finding the right balance is tricky as evolution did not design us for a long and healthy life - basically we get to the point where we have been able to reproduce and then things begin to fall apart. So the full nutrition offered by earlier diets may be very suspect even if the energy balance is good. There is probably quite a bit to learn here. Some of us (not me) are fortunate enough to have bodies that are great at self regulation. Exactly how that happens is a very interesting question.
So now we shift gears and present a few deadly recipes:-)
No one seems to be complaining about the recipes and three people have asked for more, so here are two rich desserts - a banana cream pie and a lemon tart. Both use the same shortbread crust.
I think of baking and cooking as two very different arts and suspect few really outstanding cooks are also outstanding bakers. Cooking demands knowing the state of the food and how flavors link up - there is a lot of improvisation that occurs as you cook so measurements are just a guide. Baking, at least for me, demands careful measurements for ingredients and cooking, so I use weights rather than volumes. I have a great $25 scale that measures accurately to the gram and recommend something like this to anyone who bakes. I’ll leave the conversions to you if you bake using volumes.
Both of these are very easy, delicious and something you should only eat sparingly. They were part of Sukie’s birthday present a few days ago.
The shortbread pastry crust is really simple and no rolling is required. Just put a few ingredients into a food processor, pulse a few times, and press into a tart pan. I do recommend very fresh ingredients for baking. Sugar or flour that has been opened and sitting around for a month or so pick up a funny taste.
Shortbread Pastry Crust
130 grams all purpose flour
35 grams powdered sugar
a pinch of fine sea salt (too small for my scale - I’m guessing about 1/8tsp). Note that iodized salt picks up a strange flavor when baking, so I use a fine grain un-iodized sea salt
1 stick of cold unsalted butter - about 115 grams
° heat oven to 425°F with the rack in a center position
° butter or lightly spray with a nonstick neutral vegetable oil cooking spray an 8” tart pan with a removable bottom. 9” would probably work too.
° cut the butter into chunks and put all of the ingredients in a food processor
° pulse the food processor until the pastry starts to form into pea sized clumps - about 30 seconds for me. don’t let it grow into larger pieces if you can.
° press the dough into the pan and pat as evenly as you can
° pierce the bottom of the crust with a fork to prevent it from puffing up too much during baking.
° Cover the pastry with plastic wrap and put it in the freezer for 15 minutes - this is a really neat trick that prevents the crust from shrinking when it bakes.
° Bake until the crust is golden brown - about 15 minutes for me. Cool on a wire rack. You can cover the crust and store it for a day in the refrigerator f you want.
3 large eggs at room temperature
150 grams granulated white sugar
80 grams fresh lemon juice (about 3 lemons)
a half stick of butter - about 56 grams - at room temperature - cut into chunks
4 grams of lemon zest (be careful to only get the yellow part!) I prefer a microplane for doing this
30 grams finely chopped almonds
blueberries and maple syrup - thawed frozen blueberries are fine
° In a double boiler over simmering water whisk the eggs, sugar and lemon juice together
° Cook, whisking or stirring with a spatula, constantly until the mixture becomes pale and thick - about 160°F. This takes about 10 minutes
° Remove from the heat and strain to remove lumps.
° Whisk the butter into the mixture until melted and add the lemon zest.
° Cover with a plastic sheet pressed to the surface of the lemon curd to prevent a skin from forming and let it cool to room temperature before filling the crust.
° Assemble covering the inside of the crust with finely chopped almond bits and then pour in the lemon curd. Even out with the back of a spoon or a pastry knife.
° Separate the pie from the pan - just carefully push from the bottom
° Add maple syrup to the blueberries and spoon on to individual pies of the pie when you serve. Sweeten as much or as little as you like.
Banana Creme Pie
300 grams whole milk
half a vanilla bean (you could use a good vanilla extract - maybe about a half tablespoon)
3 large egg yolks at room temperature
50 grams granulated white sugar
20 grams all purpose flour
20 grams corn starch
3 large ripe bananas
40 grams shredded coconut
whipped cream (recipe at the bottom)
° Score the vanilla bean and add to milk in a saucepan and start heating at a medium heat
° While this is cooking mix the sugar and egg yolks together. then sift the cornstarch and flour into the egg/sugar mixture to make a thick paste. Don’t let this sit around as it gets very lumpy if you do.
°Stir the warming milk. Take it off the heat when it begins to foam and remove the vanilla bean. Slowly add the hot milk to the paste whisking constantly.
° Scrape out the seeds from the bean and add to the mixture. If you use vanilla extract instead, now is the time to add it.
°Pour the mixture back into the saucepan and heat over medium heat until it begins to boil and thicken.
° Pour the mixture into a clean bowl and press plastic wrap to the top surface to prevent a skin from forming. Let it cool to room temperature.
° Slice the bananas and fold two in with the cooked mixture along with the shredded coconut. Pour into the pasty.
° Top with whipped cream and carefully release the pie from the pan side ring
120 grams heavy cream chilled
15 grams powdered sugar
1/2 tsp almond extract
° Chill a mixing bowl and beater from an electric mixer in the freezer for about 5 minutes. You can also get good results with a whisk and copper mixing bowl, but that takes some elbow grease.
° Pour the cream, sugar and almond extract into the bowl and beat until peaks form.
1 Atwater showed about 4 nutritional Calories (kcal) were available from carbohydrates, 4 from proteins and 9 from fat.
2 There has been a lot of speculation here, but now serious science is being done that supports the notion that cooking makes more energy available. A paper by Rachel Carmody and Richard Wrangham in the Journal of Human Evolution is a good example. Another study found that cooking gelatinizes starch meaning amylopectin and amylose are released and exposed to enzymes making more energy available. Cooked starches have more available Calories than those that are raw.
People have known for a long time that people who eat mostly raw food tend to be very thin compared with the general population, but serious studies on those populations are recent and the results spurred a deeper look into the more general question of food processing and available energy. Women on a purely raw food diet regularly lack the energy to have a functioning menstrual cycle even though they are eating what should be a high quality diet. The available Calories just aren’t there - we have based our food recommendations on cooked food and our bodies seem to have a body regulation mechanism that is based on cooked food - this may even be a short term evolutionary response, but that is currently speculation waiting on the verdict of science.
3 BMR is difficult to measure directly as it requires the measurement be made while the person is at rest for a long period. A more common technique is to measure a person’s resting metabolic rate (RMR). They turn out to be closely related and similar under most conditions.
A frequently used approximation is from Mifflin et. al. (1990):
BMR = 10*weight + 6.25*height -5*age +s
where weight is in kilograms, height is in centimeters, age is in years and s is +5 for males and -161 for females.
If you know your lean body mass in kilograms the Katch-McArdle approximation is probably more accurate.
BMR= 370 + (21.6 * lean body mass)
BMR in both of these approximations is in kcal (nutritional Calories)
4 One approximation is the Harris Benedict formula:
sedentary office workers use BMR *1.2 Calories
light exercise 1-3 days a week use BMR * 1.375 Calories
moderate exercise 3-5 days a week use BMR * 1.55 Calories
hard exercise 6-7 days a week use BMR * 1.725 Calories
very hard physical labor use BMR * 2.0 Calories
This is very approximate. Athletes in training can use incredible amounts. Colleen was in the 5,000 to 6,000 Calorie a day range when training - easily blowing through the approximation for very hard physical labor.
5 Normally men have higher BMRs than women even when they are the same height and weight. Men, on average, tend to have a higher percentage of lean muscle. Colleen is a bit heavier than me and, at 200 cm, quite a bit taller. She is also younger - all of this is enough to give her a higher BMR. Since she is thin and athletic you really need to use a different approximation as her body fat is low at around 10%. The Katch-McArdle formula gives a better BMR estimate of 1,925 kcal
Some unit notes: Energy is just energy. kilocalories (kcal) are the same as nutritional calories (Calories). Converting 1,925 Calories to watt-hours we see Colleen’s BMR is about 2,239 watt-hours a day. You can divide this by time to get her average power expenditure from BMR -- a bit over 93 watts.
In an inactive office worker mode she would need about 1.2 * BMR or 2,310 Calories a day - a figure that would cause a smaller person, even someone who is fairly active, to gain weight. Of course a very heavy person would have a very high BMR and their weight loss in pounds per month can be very high.
6 Note that this is energy used above and beyond your BMR. Often people assume the exercise figures are a person's total requirements during the exercise period.
7 I strongly recommend a good piece of software for logging exercise and meals.
Now I can deduce my daily requirements to be about 2,190 Calories per day. This is consistent with my calculated BMR times a factor of about 1.4 for my normal daily activity. It sort of makes sense.