A number of recent weight loss methods have been developed that explicitly recognize a close relationship between flavor and appetite. These methods include:
- Flavor-calorie dissociation as advocated by Seth Roberts in his Shangri-La Diet
- Sensory-specific satiety, as advocated in David Katz’s Flavor Point Diet
- Tastants, another approach to sensory-specific satiety, as advertised in Alan Hirsch’s Sensa Weight-Loss Program.
- Odor inhalers, a third approach based on sensory-specific satiety, as described in Alan Hirsch’s book Scentsational Weight Loss, and marketed by him as “diet pens” offered by SlimScents
At first, some of these approaches appear to be mutually incompatible. The Shangri-La theory argues that strong or familiar flavors enhance appetite when they become associated with caloric foods. The other three approaches, by contrast, claim that intense flavors or aromas suppress appetite, based upon the principle of “sensory-specific satiety”, whereby an increase in the intensity of a single flavor or odor induces satiety. However, on closer examination, all of the above theories are consistent with one another, as I will try to show. Furthermore, they each provide some useful clues about how to achieve a long term weight loss and relief from hunger cravings by paying attention to the role of flavor and other food cues. Finally, as I will attempt to persuade you, only one of the above diets is truly a type of Deconditioning Diet that can lead to long term, permanent reduction in appetite, based on the principles of Hormetism.
This post is one of the longer, more complex ones I’ve written so far. It’s almost like a full chapter in a book. For that, I apologize. But if you hang in there and try to follow the twists and turns in the science, I think at the end of this post you’ll find that certain puzzle pieces start to fit into place, leaving us with a framework that helps us to figure out some truly effective new methods of controlling appetite and weight. If you want to skip to the punch line, you can find the simple explanation at the bottom of this post, below the dotted line, under the heading “A Unified Explanation of Flavor Control Diets” and “Lessons Learned”. But for those who want to understand the science, read on….
I’d like to briefly review these weight loss programs and the underlying theoretical explanations for two reasons. First, I think that while the diets are useful, the explanations offered as to why they work are in several instances incorrect. I believe that the insulin response theory of appetite provides a more adequate explanation, and one which resolves the apparent contradiction between the two theories. Second, and more importantly, I think the insulin response theory of appetite, in combination with the philosophy of Hormetism, provides a scientific basis for the Deconditioning Diet. As detailed on the Diet page of this website, the Deconditioning Diet is a weight loss program that offers the prospect of permanent and lasting changes to appetite and long term weight loss without dependence on consuming or using specific dietary agents or devices, such as is required by all four of the above flavor control weight loss methods.
The Shangri-La Diet. I credit Seth Roberts’ book by this name for first making me aware of the connection between flavor, appetite and weight control. I tried the diet and it works. I lost 10 pounds rather easily by using Roberts’ concept of flavorless calories to suppress appetite. Roberts discovered his diet while vacationing in France. He tried some French sodas with “unfamiliar” flavors and found they suppressed his appetite almost totally. He found himself skipping meals or forgetting to eat, and he lost weight on the trip. He hypothesized that it was the unfamiliar flavors in combination with the calories from sugar that led to his appetite suppression. When he got home, he figured out that consuming sugar water with no added flavor would be as effective and simpler than using unfamiliar flavors. Consuming daily fructose water left Roberts hungry enough to eat only “about one meal every two days”, and he steadily lost about 2 pounds a week, dropping from 185 to 150 pounds in short order. He satisfied his desire for non-caloric flavors by drinking tea and chewing gum, and his desire for texture by consuming small portions of crunchy or chewy snacks like apples, nuts, or jerky.
Roberts then generalized his diet to stipulate ingesting small 200-calorie doses of flavorless calories, at least an hour before or after any meal. “Flavorless” in this case means undetectable by the nose. The sweetness of sugar appears not to count as a flavor, because it is detected only on the tongue and not by the nose. (A test of whether something is a flavor is that it will become undetectable if you pinch your nose while tasting. Sweetness, saltiness, and sourness are detected by the tongue only). So besides sugars like sucrose and fructose, Roberts found that vegetable oils work well, particularly low flavor oils like extra light olive oil (ELOO) and certain nut oils. On his website, Roberts hosts a forum where an on-line community of SLD diet followers have extended the methodology to include variations such as “nose clipping” (wearing a nose clip or pinching your nose to suppress flavors while eating regular foods) or formulating various flavorless drinks and meals. The SLD diet has clearly been successful for a great number of people, many who have found it to be the least painless way to regain control of their eating and lose a large amount of weight, with great flexibility, no restrictions on the types of food you can eat, no calorie counting, and no sense of deprivation.
Roberts’ theory of weight control can be summarized as follows: Foods with a weak flavor-calorie association will lower your body fat set point and cause you to lose weight; foods with a strong flavor-calorie association will do the opposite. He supports his theory of flavorless calories with evidence from the usually disparate fields of weight control (physiology) and associative learning (psychology). From weight control theory, he draws on the lipostatic theory of Gordon C. Kennedy. Kennedy conducted experiments with rats in the 1950s which suggested that weight or body fat is homeostatically controlled to a relatively constant “set point”. When rats were fed less calorically dense food, they first lost weight, but then adjusted by eating more of the low-cal chow to re-establish their “set point”, just as a thermostat will turn on a heater whenever the temperature of the building drops below its set point. Kennedy’s experiments were backed up by similar studies in other animals and humans. In particular, Roberts cites a series of experiments by Dr. Michel Cabanac, a physiology professor at Laval University in Quebec. Cabanac worked with human subjects, who generally find glucose water pleasant to sip. Cabanac slowly pumped glucose water directly into the stomachs of his subjects, and then had them sip sugar water. For the subjects who had sugar injected into their stomachs, the sipped sugar water became less pleasant, and they felt full. In a follow up experiment, subjects first lost 8 pounds by just eating less of their usual diet, then the glucose injection experiment was repeated. This time, the injected glucose water no longer reduced the pleasantness of the sipped sugar water, and the subjects remained very hungry. Roberts takes this to indicate that the increased appetite of this second group was caused by their actual weight being below their “set point” weight. In a further experiment, Cabanac found that letting subjects consume as much as they wanted of a restricted bland liquid diet let to significant weight loss without hunger. And one of Cabanac’s students found even faster weight loss and appetite suppression by directly injecting the liquid nutrition through a stomach tube. This suggested to Roberts that flavors raise set point, and lack of flavor (as in direct injection of sugar or food to the stomach) reduces set point.
From associative learning, Roberts draws upon the very work of Pavlov that I have cited (for different reasons) on the Psychology page of this site. He cites several experiments that appear to establish the general principle of flavor-calorie associative learning. In one study with rats, Dr. Anthony Sclafini, a psychology professor at Brooklyn College in New York, allowed rats to drink water with two different flavors. Whenever they sampled Flavor 1, a caloric starch compound (Polycose) was injected into their stomachs, but when they drank Flavor 2, plain water was injected into their stomachs. The rats developed a strong preference for Flavor 1, and this preference persisted several days even after the injections were stopped. Since they could not “taste” the injected Polycose, Sclafini and Roberts take this to provide flavor-calorie associative learning.
Putting these two lines of research together, Roberts concludes that consuming calories with familiar or strong flavors tends strengthen flavor-calorie association and preference for those flavors, and will raise the weight set point, whereas consuming flavorless calories or calories with unfamiliar flavors tends to reduce both the flavor preference and the set point. Roberts and his followers have used the hypothesized connection between flavor-calorie association and body fat set point to make predictions and extend the diet. For example, Roberts predicted and confirmed that appetite suppression and weight loss can also be achieved by consuming foods with unfamiliar flavors, so one variation of SLD is to add “crazy spices” to foods to break the flavor-calorie association, or by using nose-clipping to suppress detection of flavors while eating.
However, the success of the SLD diet in itself does not prove that Roberts’ explanation for why it works is correct. I believe that Roberts explanation is incorrect for several reasons: (1) the set point theory is not empirically testable; (2) Roberts learning theory is based on a misunderstanding of Pavlov’s theory of associative learning; (3) Roberts’ learning theory makes several false predictions and fails to explain other facts about weight loss; (4) there is an alternative explanation for why SLD works that can better explain these other facts about weight loss. I’ll take up these criticisms in order.
Lipostatic set point theory of weight control. According to Roberts, if you weigh less than your set point, “you will be hungry and think about food” and the bigger the gap, “the more you will think about food, and the more food it will take to feel full when you eat. It is nearly impossible to weigh much less than your set point for a long time–the hunger becomes unbearable.” (SLD, p.9) Similarly, if you weigh more than your set point you will not be hungry and “When you eat, you will feel full rapidly.” It is apparently not your weight, but your body fat content, that is compared with a set point. “When your body-weight regulatory system detects that you have less fat on your body than your set point, it makes you more hungry than usual between meals and increases how much food you need to eat to feel full.” (SLD, pp. 41-2)
And yet the lipostatic set point does not seem to operate like temperature set point on a thermostat. Whenever the temperature is below the thermostat set point, the heater turns on immediately, or after a short delay. And whenever the temperature is above the set point, the heater never turns on–until temperature drops on its own, or with the assistance of air conditioning. But this is not the case with body weight “set point”. When you weigh 8 pounds less than your set point, you don’t eat continuously in one session until the set point is reached. Likewise if you weigh 8 pounds more than your set point, you don’t totally stop eating until your weight drops to the set point level. Instead, most people have a certain rhythmic frequency of eating alternating with not eating. Something else regulates this rhythm of meals even when we far from set point. (As I’ll suggest below, that “something” is the hormone insulin).
Roberts acknowledges that set point is not fixed, and he posits that your set point is directly influenced by what you eat. He explains that between meals, your set point goes down gradually, and the higher your set point, the faster it goes down. But, according to Roberts, your weight goes down faster than your set point, which is “why not eating causes hunger–and why diets that deprive you of food don’t work.” Eating foods with flavor-calorie association will immediately raise set point; and the stronger the association, the greater the increase in set point.
But this is an odd sort of set point that is never really set! The very thing that is controlled by set point (eating) itself causes the set point to change. It is as if a thermostat setting were itself influenced by the temperature in the house, rather than by the intentional decision of the house occupants. The set point concept starts look very flexible, with a lot of external influences and circular feedback loops. (It reminds me of how epicycles were added to the Ptolemy’s geocentric theory of the solar system in order to explain the apparent zigzagging of observed orbits of plants and stars around the earth, until Copernicus eventually came up with the simpler heliocentric theory in the 16th century). If set point is always changing, how could you verify this? It just seems like a convenient theoretical entity that can never be proven or disproven. Roberts’ claim that highly flavored caloric foods cause weight gain because they are “high set point foods” is an example of this circular reasoning.
There is also one other weird aspect of the lipostatic theory of weight control. It seems to commit us to the view that each of us has a “natural” set point weight or fat mass, which we cannot change. Unless, that is, we following the Shangri-La Diet and we commit ourselves to eating bland foods, at least periodically–forever. If we are born with a “fat” set point weight and eat flavorful caloric foods, we are destined to get fat. Naturally thin people who can eat flavorful foods are just lucky, because…they are born with a low set point. Somehow, I doubt this.
This does not mean we should throw out the concept of a set point. We just have to pick a physiologically verifiable set point. True examples of set points and homeostasis are blood pH and body temperature which are controlled in a narrow range, under normal circumstances. There is not anything that you can do to alter these set points, and nor would you want to do so. Fevers, hypothermia, and ketoacidosis are indicators of disturbance or pathology, and the organism works hard to restore temperature or pH to the healthy set point values in such cases. Furthermore, the physiological basis of these homeostatic mechanisms for regulating body temperature and blood pH are well known. So I think that extending the set point concept to such a variable quantity as body weight is neither legitimate nor useful, and there burden of proof should be on its advocates to demonstrate an empirical basis for it.
Roberts does acknowledge physiology when he claims that set point is related to the concentration of leptin in the in the bloodstream:
In order to regulate the amount of body fat, the brain must be able to know how much body fat you have, just as a thermostat needs a built-in thermometer to keep track of the room temperature. Leptin serves as the brain’s body-fat thermometer: The concentration of leptin in your blood tells your brain how much fat is in your body. Leptin is produced by fat cells. When your body fat goes up, so does the amount of leptin in your blood. When your body fat goes down, leptin goes down. (SLD, p. 143)
This is getting closer to a bona fide physiological explanation. In fact, I agree that hormones drive weight control, I just think that Roberts has chosen the wrong hormone. There are two problems with choosing leptin. First, while leptin does play a key role in signalling satiety to the brain, it is not the whole story. First, the leptin signal does not always work. People can become resistant to leptin. The brain can’t always “hear” leptin even when it is elevated, leading people to overeat. This is especially true with obese individuals. The leptin signal can also be blocked by ingesting a meal high in carbohydrates, since carbohydrates lead to high triglyceride levels in the bloodstream, and this can block the leptin signal, again leading to overeating. Furthermore, leptin is not the only player in hunger signaling. The brain integrates the signal from leptin with that of other appetite signallers, including hormones such as ghrelin, and neurotransmitters such as neuropeptide Y, serotonin, dopamine and various opioids. We need to look further up the stream to the root cause of hunger, not the “messengers”.
Another problem with the leptin theory is that is inconsistent with at least two observational facts:
- The observed variability in body weights in the population. Leptin is known to increase in fat individuals and decrease in lean individuals. So lean individuals, who have lower levels of leptin, should be hungry and want to eat more, until their leptin reaches a “satiating” level. However, that would seem to rule out the existence of lean people. Everyone should eat until their leptin level reaches the “normal” level. And yet that is absurd, because most lean people are not ravenously hungry. That implies that every individual can have a different “normal” leptin level. That is in fact the case, but then it removes leptin as a homeostatic regulator of weight or body fat.
- The suppression of hunger observed during extended fasting. After an initial period of cravings, light-headedness or other hypoglycemic response, people who are fasting for extended periods of 12-24 hours or longer commonly report appetite suppression and a surge of energy. This cannot be explained by the lipostatic set point theory or Roberts version of it. To repeat what I wrote above, according to Roberts, your weight goes down faster than your set point, which is “why not eating causes hunger–and why diets that deprive you of food don’t work.” But if weight is dropping faster than set point during fasting, your hunger should increase monotonically, that is, it should get steadily more intense, without dissipating. Yet that is not what happens. After a period of adjustment, as insulin drops, glucagon rises along with adipose tissue lipase, and adipose tissue begins releasing free fatty acids into the bloodstream, while the liver breaks down glycogen into glucose and proteins into glucose via gluconeogenesis. So a constant level of glucose in the bloodstream is restored, and there is no sense of deprivation or “unbearable” hunger. Most people who fast, including myself, find it quite pleasant. I think the bad rap on fasting comes from people who have not studied the physiology of fasting and either haven’t tried it, or haven’t given their bodies enough time to adapt.
The glucostatic theory. There is a much better alternative to the lipostatic theory of hunger and weight regulation: namely, the glucostatic theory of Jean Mayer, as it has been developed into a fuller theory of the homeostatic regulation of blood glucose by the hormone insulin. While there are many hunger signaling compounds, ultimately these signals are either responding to or predicting what is happening or likely to happen to blood glucose concentrations. In the glucostatic theory, there is also a set point that controls hunger, and it is blood glucose concentration. According to the glucostatic hypothesis, there is no body fat “set point” that our body attempts to defend, no “natural” level of fatness or leanness. This is evidenced by the fact that it is possible to gain or lose a lot of weight and main these changes. Whether we are fat or lean, our body strives to maintain glucose concentration within a physiologically tolerable range of about 70 to 150 mg/dL. Below that level, hypoglycemia sets off hunger pains, headaches and–in extreme cases–shock and coma. Above that level, there can be both acute problems with high blood glucose and chronic problems with elevated blood glucose, leading eventually to hyperinsulinemia and diabetes. There is an excellent discussion of the glucostatic and lipostatic theories of weight control in Chapter 24 of Gary Taubes’ Good Calories, Bad Calories, delineating further problems the lipostatic theory.
This is not to say that weight is not remarkably stable in most people for long periods of time. But this stability is not the result of homeostatic control. Rather than being a “set point”, body fat may be a “settling point”. Taubes is very articulate on this point:
Life is dependent on homeostatic systems that exhibit the same relative constancy as body weight, and none of them require a set point, like the temperature setting on a thermostat, to do so. Moreover, it is always possible to create a system that exhibits set-point-like behavior or a settling point, without actually having a set-point mechanism involved. The classic example is the water level in a lake, which might, to the naive, appear to be regulated from day to day or year to year, but is just the end result of a balance between the flow of water into the lake and the flow out. When Claude Bernard discussed the stability of the milieu interieur, and Walter Cannon the notion of homeostasis, it was this kind of dynamic equilibrium they had in mind, not a central thermostatlike regulator in the brain that would do the job rather than the body itself.
This is where physiological psychologists provided a viable alternative hypothesis to explain both hunger and weight regulation. In effect, they rediscovered the science of how fat metabolism is regulated, but did it from an entirely different perspective, and followed the implications through to the sensations of hunger and satiety. Their hypothesis explained the relative stability of body weight, which has always been one of the outstanding paradoxes in the study of weight regulation, and even why body weight would be expected to move upward with age, or even move upward on average in a population, as the obesity epidemic suggests has been the case lately. (GCBC, pp. 428-9)
In effect, the apparent stability of weight results from a relative constancy in many factors–hormone activation, food intake, physical activity–any change in which can shift this balance to a new stability point.
The glucostatic theory also provides explain how appetite can change on a minute-by-minute basis. Typically, a drop in blood sugar is associated with hunger and the initiation of eating. While blood sugar can drift down on its own, the more usual scenario is that food cues such as aromas, or the time of day, lead to a psychologically triggered secretion of insulin are the immediate causes of a dip in blood sugar, with its attendant hunger signals. This usually occurs at meal times because of prior conditioning. Blood sugar can also drop when the body is running low on glycogen and is having difficulty switching over to fat or ketones, or when blood sugar is depleted by exercise or other activity that outstrips the ability to resupply glucose from the tissues.
Unlike blood glucose and insulin–which can change by the minute and correlate well with the well known rhythms of appetite and satiety–body fat and body weight change much more slowly, and therefore do not seem to be good candidates for direct control, as required by the set point theory of weight regulation.
Associative learning. There is a second problem with Roberts’ explanation of why the Shangri-La Diet works. One of Roberts central claims is that body fat set point is influenced by flavor-calorie association, and that the strength of the association between flavor and calories comes about by means of associative learning. In this regard, Roberts explicitly invokes Pavlov to explain why the the SLD requires that flavor and calories be detected at separate times:
In Pavlov’s experiments, the bell was the signal and the food was the outcome. The food was given at the same time the bell was turned off. Had the food been given many minutes after the bell was turned off, the dog would not have associated them at all. (SLD, p. 47)
Roberts further supports this contention about timing by citing experiments with rats indicating that “a flavor-calorie association was learned even when there was a thirty-minute gap between eating the flavor source and eating the calorie source” (p. 70), but a one-hour gap was sufficient to prevent such a learned association from forming. The flavorless calorie window of one hour stipulated for the SLD is based on the Pavlovian principle that “the more delayed the outcome, the weaker the association”.
But I believe Roberts’ interpretations and conclusions regarding Pavlov’s finding are not correct here. First, Pavlov did in fact show in his experiments with dogs that a delay between a stimulus such as a buzzer, and a delayed feeding would merely reinforce a delay in the salivation. The dogs would wait the required amount of time:
As his technique became more practiced, Pavlov’s laboratory began investigating the canine sense of time. After a dog was trained to salivate at a flash of light, the delivery of the stimulus was postponed by three minutes. Before long, the dog learned to anticipate the delay. Three minutes after the signal, the animal’s mouth would water. (The Ten Most Beautiful Experiments, p. 132).
But second, note that Roberts in the previous quote is talking about an association between the bell and the food presentation (two stimuli). That is indeed associative learning, because both the bell and the food can be perceived. The bell and food repeatedly occur together, and become psychologically associated. This association is strengthen considerably by reinforcement. The bell is a conditioned stimulus (CS) and the food is an unconditioned stimulus (US) and classical conditioning is all about associating CS with US, by virtue of reinforcing the unconditioned response (UR). In this case the UR is salivation and an increase in appetite, and the reinforcement is the rise in the dog’s blood sugar that occurs after it eats the food. (If the food was taken away before being eaten, there would be no reinforcement).
Yet in his explanation of why the Shangri-La diet works, Roberts talks about an “association” between flavor and calories. However it is a misapplication of the concept of associative learning to speak of “associating” a perceivable entity (flavor) with an unperceivable physiological reaction (detection of “calories”). The calories cannot be directly perceived by the sensory apparatus. Rather, the ingestion of food results in a physiological response, a rise in blood sugar, and its further consequence, secretion of insulin. There is a perceivable consequence of the calories (hunger), which is possible to associated with certain stimuli. But that is not what Roberts is claiming. He says that we learn to associate flavors with calories, not with hunger or satiety. The relationship between a perceivable stimulus and a physiological response, if reinforced, gives rise to classical conditioning. This is distinct from associative learning, in which two perceivable stimuli become associated with each other when they occur together repeatedly.
Put another way: it is not strictly correct to say that people “learn” to make “associations” between flavors and calories, especially since they are not directly aware of the calories (other than by reading food labels). All we can discern is whether or not the flavorful food provides relief from hunger. Roberts sometimes seems to conflate awareness of calories–a requirement for associative learning–with physiological detection of calories. But physiological “detection” by the digestive system is not associative learning.Strictly speaking, it is not a flavor-calorie association, but rather an association between flavor and the satisfaction of appetite. An even better way is to formulate this as a stimulus-response relationship, since the primary response is physiological and not conscious. In this case direct response to the stimulus of flavor is insulin secretion, and indirect responses are blood sugar and appetite. Pavlov primarily studied the conditioning of stimulus-response relationships; the associative learning was merely an explanation for how a secondary, conditioned stimulus could become secondarily associated with the primary unconditioned stimulus.
Even if we were to accept Roberts position that unconscious detection of caloric foods by the digestive system is sufficient for calories to play a role in “associative learning”, we are left without an explanation of how calories per se can be detected. In this regard, the digestive system does not recognize all “calories” as a monolithic unit of food, as Roberts seems to suggest. His explanation of how low carb and “good carb” diets work overlooks significant differences in how calories in different types of macronutrients interact with the digestive system. He seems to assume the only difference between carbohydrates, proteins and fats is how quickly they are released and detected:
When a food is digested more slowly, the calories in that food are detected more slowly. Thus there is more of a gap between the signal (flavor) and the outcome (calories). I believe this is why low-carb and good-carb diets work: They replace foods that are digested quickly, such as bread, with foods that are digested slowly, such as vegetables. The foods that are digested more slowly have weaker flavor-calorie associations. They raise your set point less. (SLD, p. 47)
But this does not seem very plausible. Low molecular weight oils can be digested and absorbed as least as quickly as most starches. The reason that oils lead to better appetite suppression and weight loss than simple carbs is not that they are detected more slowly, but rather than they induce little or no insulin response. Furthermore, fats will result in increased appetite in weight if they are combined with even a modest amount of carbohydrates, because the insulin response will cause them to become “fixed” with the glucose to form triglycerides in the adipose tissue. It is critically important to consider the types of macronutrients and how the endocrine and digestive system responds to them alone and in combination.
A better explanation for Robert’s one-hour rule is based on the dynamics of insulin secretion, and the way that this can become conditionally mediated via sensory stimulation of the vagus nerve . When correctly understood in terms of insulin response, the one hour rule should be changed to an asymmetric rule. In fact, one should only need to wait about 15-30 minutes after ingesting non-caloric flavors before consuming calories, but depending on the meal size, one may need to wait an hour or more before consuming flavors after eating. This is because the pre-prandial insulin response is shorter and smaller than the post-prandial insulin response.
In fact the one-hour rule must be differentiated even further, based on the macronutrient composition of the calories, and it leads to a number of predictions that diverge from the SLD:
- Wait at least 15-30 minutes after a flavor or aroma before ingesting carbohydrate containing foods. (This is to allow the pre-prandial insulin response to subside, and blood sugar to renormalize).
- Wait at least an hour after consuming flavorless carbohydrate-containing foods before ingesting flavors. (This is because the post-prandial insulin is much larger and takes longer to return to baseline than the transient flavor-induced pre-prandial insulin response).
- The same rules apply to meals which contain large amounts of protein. Protein is much less insulinogenic that carbohydrates, but large protein meals and certain types of protein are insulinogenic.
- You can eat fats and flavors together without any worry, because fat is not insulinogenic alone–it requires the presence of a little carbohydrate or a lot of protein to be removed from the bloodstream by insulin.
The flavor-insulin response, which is mediated by the tractus solitarus in the brain and the vagus nerve, is a conditioned response. It will be strengthened whenever the flavor cue is concurrent or closely followed by a the ingestion of insulinogenic foods, foods which in themselves produce an insulin response when detected by the glucose receptors in the stomach and intestines. Essentially, the flavor-insulin response is a predictive response that readies the digestive tract for food that is coming, by making the food more absorbable. Secreting pre-prandial insulin for a pure fat meal or a small protein meal has no value, and that conditioned response will tend to extinguish. So that leads to another set of predictions. In particular, despite what Roberts claims, fats and sugars should work very differently in the SLD. So here are some predictions made by the insulin-regulated glucostatic theory of hunger, all of which are either not predicted by SLD or are diametrically opposed to what SLD would predict:
- Consuming pure fats like olive oil or heavy cream should suppress appetite even if they are flavored. This only works, however, if no more than a trace of carbohydrates or proteins. (This works because fats are noninsulinogenic).
- Increasing the dose size of oils even to large doses should increase appetite suppression
- Increasing the dose size of sucrose or glucose beyond the minimum dose, should reduce appetite suppression. A small amount of sucrose or glucose increases satiety because it raises blood sugar slightly, and flies “under the insulin radar”. Insulin is not secreted until blood sugar rises above a certain threshold, typically 120 mg/dL or so. But once it exceeds that threshold, insulin kicks in, and blood sugar drops.
- Sipping sucrose or glucose slowly will maintain appetite suppression. If this is done slowly enough, the addition to blood sugar just balances out the amount of blood sugar consumed to meet metabolic needs. But this is a careful balance.
- Increasing the dose of fructose, xylitol, erythritol, or other non-insulinogenic sugars should suppress appetite at any dose.
Prediction 4 is consistent with SLD. Predictions 2 and 5 are consistent with SLD, but not predictable from it. Predictions 1 and 3 are contrary to the fundamental assumptions of SLD, and would not be predicted by it.
For anyone who is interested in subjecting SLD to a test, I would be interested in their experience in attempting to verify or refute the above predictions, especially 1, 2 and 3.
The Flavor Point Diet. (FPD). This diet, created by David Katz, is based on the concept of “sensory specific satiety”. Eating a meal with flavors in multiple “competing” categories such as sweet, salty, or savory, somehow stimulates the “appetite center” of your brain, causing overeating. Limiting the flavor categories in a meal or snack to one or two flavor types makes it easier–more rapidly and with less food– for your brain and stomach to reach the “flavor point”, a state of satiety that causes you to stop eating, with less food. This concept is reflected in the popular notion of “multiple stomachs”, whereby you can feel stuffed after eating one course of a meal, but you often seem to find additional appetite for dessert or something different. The FPD advises you to restrict the number of simultaneous flavors within a meal, but overcomes the potential boredom by allowing you to eat a variety of flavors over the course of a day or week–just not “excessive variety” all at one time.
Katz bases most of his case for sensory specific satiety on the way in which flavors stimulate the production of neurochemicals that activate the hypothalamus:
As soon as you taste food, the sensory information registers in the hypothalamus in the brain, which, depending on the flavor of the food, sends out signals to eat more or less. Because of this sensory relay system, the appetite center in your hypothalamus can become aroused–and in some cases overly aroused–by how a food tastes. (FPD, p. 4)
As soon as you bite into any food, sensory stimulation of nerve endings on the tongue leads to the release of a number of chemicals, including opioids, into the bloodstream. You release more opioids–the body’s natural version of drugs like morphine–when you consume foods high in sugar and fat, creating a powerful, neurochemical drive to overeat those foods. These opioids and other chemicals enter the bloodstream and carry their messages to the hypothalamus, which sends out yet another set of chemicals to regulate appetite. The more flavors your taste buds register, the more stimulated the hypothalamus becomes, releasing the hunger-promoting neuropeptide Y. When you taste a lot of flavors at once, the brain releases a lot of neuropeptide Y. Meanwhile, in response to the smell and taste of food, your stomach produces the hormone ghrelin, which also stimulates appetite. It continues to produce this hormone until you eat enough food to literally fill your stomach and stretch the stomach wall. Farther down the line, in your intestines, levels of several hormones rise to varying degrees–depending on the nature of your meal–either inducing more hunger or turning off hunger. (FPD, p. 4-5)
Katz claims that by “organizing” the flavors in our diet, we can manipulate this chain of chemical and signals and “subdue the appetite center in your brain sooner, before you’ve overeaten.” (p. 9) Katz also does acknowledge a role for insulin in controlling blood sugar, and points out that fast carbs cause a rapid blood sugar surge and an insulin spike which tends to overshoot and lead to a drop in blood sugar, whereas low glycemic carbs like oatmeal result in a lower rise in blood sugar, a slower release of insulin, “no rapid surge and dip in blood sugar levels” and sustained satiety. But he does not make any direct connection between flavor and the insulin response, putting the onus on the neurochemical triggers like opioids, neuropeptide Y, and ghrelin.
The FPD establishes how flavors can begin a cascade that induces appetite. Katz is probably correct that the effects of individual flavors have a saturating effect on this response, and that the response can be increased by combining multiple flavors. I think that Katz overstates the role that neurotransmitters and leptin play in hunger and appetite. Neurotransmitters and leptin are are important as primarily signaling compounds to the brain, but they are not the primary causal agent in that chain. Insulin is much more directly involved in the control of appetite, because it is insulin that reduces blood sugar to physiologically unsustainable levels in the first place, and the signaling compounds are merely the messengers. Blaming these signalling compounds for hunger is like blaming a witness for the crime. Perhaps neuropeptides and leptin can signal hunger without concurrent insulin response. I think this is unlikely, but if it occurs, I suspect it is because these signals have independently become classically conditioned to respond to flavors and other food cues that have become associated with the presence of food. As with insulin, these neurotransmitter responses to sensory stimuli can be deconditioned upon experience. However, Katz treats the neurochemical responses to flavors as “hardwired”, overlooking the fact that they are learned or conditioned responses. The responses will strengthen if reinforced by eating food that increases blood sugar, and will weaken if not reinforced. His claim that sugar and fat alone cause the release of opioids into the bloodstream is not documented and seems unlikely. The tongue does not detect sugar and fat directly. In fact, some neurological research by Woolley et al at UCSF indicates that it is flavors, and not the macronutrient content of foods, that stimulate opioid secretion by the nucleus accumbens in the brain. To the extent that the hypothalamus is engaged, it requires a detectable signal such as flavor. As Teff showed (see Diet page), the tractus solitarus will trigger the vagus nerve to secrete insulin only in response to flavor and scents that it has learned and expects to be associated with blood-sugar raising foods.
Sensa tastants. Closely related to the Flavor Point diet is the use of tastants as Alan Hirsch has developed for his Sensa Weight Loss Program. The Sensa tastants are intense non-caloric flavors that are sprinkled on foods to enhance flavor intensity. The tastants are matched the flavor class of the foods: savory tastants are sprinkled on savory foods, while sweet tastants are sprinkled on sweet foods. The principle of sensory-specific satiety is identical with that of the Flavor Point diet. One is still advised to eat only one or two “flavor categories” of food at a meal. The advantage of Sensa over FPD, perhaps, is that the satiating “flavor point” is reached earlier in a meal, with less food and fewer calories consumed than otherwise. It seems to me the the same result could be achieved by spicing your savory foods and adding non-caloric sweeteners to your desserts.
As with FLP, however, tastants will do nothing to fundamentally alter or extinguish strong flavor-insulin responses. So one remains vulnerable to a strong appetite returning when combining foods, and has to observe the principle of limiting the number and variety of flavors at any given meal.
The bottom line, with both FLP and Sensa, is that by confining yourself or intensifying a single flavor, you allow the insulin response to that specific flavor to saturate earlier, thereby limiting the appetite-inducing property of insulin. If you were to add more successive different flavors, you will tend to stimulate separate flavor-detection pathways and add another wave of insulin secretion and stoking of appetite.
Odor inhalers (Scentsational Weight Loss). Direct exposure to saturating levels of food aromas is yet another way to exploit the sensory specific satiety mechanism. Even though it exploits the same mechanism, it is in an entirely different class and–as I will argue–odor inhalers have the potential to engender long term appetite deconditioning and weight loss.
Alan Hirsch describes his discovery of this phenomenon in his book Sensational Weight Loss (SWL), and the same concept is exploited in the commercial “diet pens” sold by SlimScents. There are a number of other related pens and inhalers marketed as “aromatherapy”. Hirsch reports a remarkable conclusion from his six month study of the use aroma devices for to quell appetite and spur weight loss. The study included 3193 participants, 86% of whom were women, and most of whom were significantly overweight (average weight = 217 lb.). Participants were given small odor inhalers, similar to lipstick dispensers, containing pleasant smelling substances. Three aromas were used: green apple, peppermint, and banana. Hirsch found that varying the aromas was more effective than sticking with the same aroma. Participants were asked to open and sniff the aromas whenever they got hungry–three sniffs in each nostril. But otherwise there were no forbidden foods or other dietary restrictions. The results were impressive. The average weight loss was 5 pounds per month. Some people lost more than 100 pounds over the six month study. What is especially interesting to me, however, is that there was a permanent deconditioning effect, along the lines of the Deconditioning Diet (see Diet page). Here are Hirsch’s comments regarding “deprogramming” of the participants’ learned responses to flavors, leading to long-term changes in eating habits:
Just as we have learned to respond to the smell of certain foods by feeling hungry and wanting to eat, we can, in a very real sense, “unlearn” or deprogram ourselves. For many people the smell of any food triggered hunger. Smell a doughnut, salivate; smell a pizza, and the stomach growls. Actually, most of us experience this much of the time; but in the overweight person, this conditioned, or learned, response can be quite. powerful. It’s exciting to realize that people can recondition themselves to smell an odor and not respond with hunger. In the absence of a food associated with the smell, hunger disappeared, the desire to eat subsided, and a pattern was broken. In many cases, this was a long-standing pattern that was broken during the six month study. (SWLP, p. 33-4)
Further evidence of a long term change was reflected in the frequency of use of the odor devices during the study. At the beginning of the study, participants reported needing to sniff 200 times a day or more; by the end of the study, only occasional sniffing was needed to keep appetite in check. So the the aroma devices don’t themselves become habit forming–the dependence actually decreases over time as new eating habits are consolidated.
In Chapter 4 of Scentsational Weight Loss, Hirsch provides some additional advice, namely that we pre-saturate our satiey centers by sniffing our meals before the first bite to letter the odor molecules fully stimulate the olfactory bulb. He recommends that we slowly chew and savor the flavor of each bite in order to “fool” the hypothalamus into “believing that more food has been ingested than is actually the case. He also recommends eating food warm or hot to maximize the aroma, adding spices whenever possible, and choosing the more strongly flavored versions of foods. “For example, if you eat popcorn, eat the cheese-flavored varierty, or choose an onion bagel over a plain one.” (SWLP, p. 62).
On the face of it, this advice is diametrically opposed to that of Seth Roberts’ Shangri-La Diet! Roberts argues that blander foods induce weight loss, whereas Hirsch is arging for more intensely flavored and spiced foods. Who is right and what gives?
Here is where all three diets — SLD, FPD, Sensa, and SWL– come together. After promoting the enhancement of flavor, Hirsch adds:
Try limiting your food choices at any one meal. We still encourage variety, but not necessarily all at one time. We have found that people who want to lose weight should eat only two or three different foods during a meal rather than eating a little bit of many foods…Avoid buffet tables and “all you an eat” food bars. Even salad bars can be dangerous…Even with a smell device, too many selections and unlimited choices spell potential overeating. (Sensa Weight Loss Program, p. 63-64)
A Unified Explanation of Flavor Control Diets. Here is my summary explanation that can account for the observed effects of all four dietary approaches: Familiar flavors can induce a pre-prandial insulin response which leads to increased appetite and weight gain. This preprandial insulin response saturates separately for each basic flavor type (savory, salty, sweet, etc.). This saturating pre-prandial insulin response is is a learned response, and one that is reinforced only if it results ultimately in a rise in blood sugar and psychological satiation. If that flavor or aroma is not followed by more eating, it will eventually diminish or extinguish as a cue.
Here then is how this explains each of the four flavor control diets:
- SLD explanation. Eating foods that contain carbohydrates (and to a lesser extent, proteins) with strong, familiar flavors will lead to a rapid pre-prandial insulin response which may be enough to cause a dip in blood sugar, stoking hunger and leading to further eating. Eating a small amount of bland or flavorless carbohydrate will satisfy hunger by slightly raising blood sugar, and will induce only a small post-prandial insulin response, insufficient to cause a rapid decline in blood sugar, so appetite will remain suppressed. While SLD keeps insulin in check, it does so only so long as flavor cues are not present. But SLD does nothing to weaken the connection between familiar flavors stimuli and their insulin response, it merely eliminates the stimuli. One would expect appetite to return when the familiar flavors are re-introduced.
- FPD explanation. Eating foods with familiar flavors will induce two insulin responses: a small pre-prandial response and a larger post-prandial response. However, the pre-prandial insulin responses for that flavor peaks within about 4 minutes after exposure, and sensory detection of that flavor will rapidly saturate. After it saturates, any further exposure to that specific aroma (or aroma class) will not induce any further insulin response for an extended period of time — up to about an hour. However, introducing new flavors, aromas or other food cues will cause additional secretion of insulin, increasing appetite. The more intense the flavor or aroma, the faster the saturation.
- Sensa explanation. The explanation for Sensa is the same as that for the FPD. Activation of a sensory response to a flavor will induce a small and rapid pre-prandial insulin response. For a single flavor or aroma class, the detection of that flavor will saturate after a certain amount of time, after which it will not prompt any further insulin response. If multiple aromas are sniffed, then all those smell receptors will become saturated, after which any further pre-prandial insulin response will become muted. Since the Sensa tastants contain no or minimal calories, they provide a way to reach flavor saturation faster, with fewer calories. In that way, the use of tastants is more effective than the Flavor Point Diet, since fewer net calories are consumed while satisfying one’s appetite.
- SWL (odor inhalants) explanation. With sensory-specific satiety approahes like FPD and Sensa, the stimulus-response relationship between flavor and pre-prandial insulin response is reinforced. The stimulus may saturate, but the connection with the insulin response remains in place, so that at subsequent meals, the flavor will induce pre-prandial insulin. However, with SWL, the relationship between flavor and pre-prandial insulin is not reinforced, so it extinguishes! This is a crucial difference. For this reason, SWL is actually a deconditioning diet that results in long term changes in the flavor-insulin response that suppress appetite and lead to weight loss. Eventually, the aroma inhalers are no longer needed, or only rarely, to maintain the “deprogramming” that Hirsch alludes to. By contrast, SLD, FLP and Sensa are not deconditioning diets, but merely methods of suppressing or limiting appetite that work by either minimizing or saturating the stimulus of food cues, but doing nothing to weaken the flavor-insulin response.
Lessons learned. Where does this leave us? There is a lot to be learned from all four flavor control diets. Flavors and aromas that become associated with foods, particularly carbohydrate-containing foods, strengthen the insulin-response to those flavors and aromas, increase appetite, and tend to increase the consumption of those foods, leading to weight gain. This insulin response and the resulting appetite can be significantly dampened by limiting the variety of flavors while eating. However, a much larger benefit is possible by using flavors to decondition your appetite:
- You can expose yourself to flavors without carbohydrates or other insulinogenic foods and this will dampen the flavor-insulin response, and lead to a decrease in appetite that is induced by food cues.
- By exposing yourself to a variety of flavors or aromatic stimuli without eating, you will saturate a fuller range of satiety centers and even more effectively block an insulin response. Sniffing a variety of aromas of different types–savory and sweet– without eating can be helpful in curbing appetite.
These findings have been incorporated into the second phase of the Deconditioning Diet, as described on the Diet page in this website.
Please leave a comment. You can also check out (and maybe start) a discussion on the Diet Forum linked to this blog.