Physiological Regulation: How It Really Works

Abstract

Contrary to dogma, much physiological regulation utilizes learning from past experience to make responses that preemptively and effectively neutralize anticipated regulatory challenges. Understanding physiological regulation therefore requires expanding explanatory models beyond homeostasis and allostasis to emphasize the prominence of conditioning.

Introduction: Homeostasis and Allostasis.

Claude Bernard is credited with describing the basic tenets of physiological regulation. Recognizing the importance of the body’s being able to maintain a stable internal environment, one that enables vital processes to function optimally during times of environmental fluctuations and perturbations, Bernard envisioned an armamentarium of reflexes through which the body is able to detect, and subsequently make responses to counter, harmful deviations in one or another internal variable whose level is critical to sustain life (Bernard, 1870/1973). Thus, Bernard’s milieu intérieur was able to remain relatively stable, protected from environmental challenges by diverse regulatory responses that allow an individual to thrive and reproduce. Bernard’s list of regulated physiological parameters was short, encompassing oxygen and energy supply to tissues (e.g., blood glucose), body temperature and several cardiovascular parameters (blood volume, pressure and osmolality), what today would collectively be known as vital signs in a medical setting.

The underlying mechanism purported to account for regulation in Bernard’s schema was negative feedback. When a perturbation occurs that elicits an increase of blood glucose or a decrease of body temperature, regulatory reflexes are initiated to counter the change and bring the level of the impacted parameter back to a safer, more optimal level. Walter Cannon popularized this view of regulation, expanding it to numerous other parameters in the body, and coined the term homeostasis to describe the continuously ongoing processes that function to maintain optimal levels of parameters necessary to sustain life (Cannon, 1929). Cannon’s concept of homeostasis was incredibly influential and dominated consideration of regulation for most of the last century. Subsequent to Cannon’s writing, two important transformations occurred that blur the image and interpretation of homeostatic regulation. The first was the application of engineering control theory to physiological systems (Wiener, 1948), inserting terms such as error signals, set points and central controllers, concepts that we, like many others, believe do not appropriately apply to physiological regulation, and that have misled several generations of scholars (Kanosue et al., 2010; McAllen et al., 2010; Partridge, 1982; Ramsay and Woods, 2014; Romanovsky, 2007). The second was the gradual expansion of the scope of ‘homeostasis,’ to the point that it is now commonly applied to describe, or account for, any change to a system whether at the molecular, cellular, systems, individual or even social or population levels.

A more recent development in considering regulation has been the widespread recognition that homeostasis, as defined by Cannon, fails to account for common regulatory phenomena including the role of experience and the absence of a rigid, invariant set point (Ramsay and Woods, 2014). These shortcomings have been the topic of considerable debate, and myriad alternative models have been put forth to better account for regulation [e.g., such as ‘homeostasis’ but with an adjustable set point (Hammel, et al., 1970; Gordon, 2009)]. The model that has gained the most traction is allostasis, which in its original form differed from homeostasis by formally positing that the defended level of a vital parameter can in fact change under some circumstances (i.e., there is no invariant set point), and that an individual can benefit from past experiences to better anticipate future challenges to the internal milieu (Sterling, 2012; Sterling and Eyer, 1988).

Over the years since its original description, allostasis, like homeostasis, has been considerably transformed by its usage in the literature. Thus, a commonly accepted contemporary view is that homeostasis refers to regulation that occurs under normal or physiologic situations in which the body effectively counters whatever challenges occur. The implication is that similar challenges were present in the evolutionary history of the organism, allowing efficient reflex compensatory responses that counter them to evolve. Allostasis, on the other hand, refers to a dysfunctional regulation that occurs when confronted with artificial or pathophysiologic situations, perhaps situations that never occurred in evolutionary history. These odd situations or environments create an ‘allostatic load’ that reflects the added costs resulting from abnormal or inefficient regulation (McEwen, 1998) and reveal the fragile side of our robust regulatory systems (Csete and Doyle, 2002; Kitano et al., 2004). Common examples include the results of exposure to calorically-dense, highly-palatable foods, to drugs of abuse, and/or to unrelenting psychosocial stress. Thus, allostatic regulation is now often invoked to account for the rising prevalence of obesity/diabetes, drug addiction and depression in these situations, respectively, and all are considered allostatic states in which maladaptive and inefficient responding to the complex challenges presented by modern society lead to pathology. More nuanced contemporary views of homeostasis and allostasis in physiological regulation explain the growing recognition that there is no basis for invoking a central controller that purposefully coordinates responses to attain a desired set point and instead are more compatible with a balance-point view of regulation (Ramsay and Woods, 2014).

Mechanisms of Physiological Regulation: Negative Feedback.

Homeostasis/allostasis aside, what do we actually know about physiological regulation? Both Bernard and Cannon wrote that regulation occurs as organisms react to physiological insults or challenges, often due to environmental changes or pressures. When body temperature goes down in a cold environment, heat-generating reflexes are elicited; when blood glucose goes up after a large meal, glucose-lowering reflexes are elicited; and so on. This type of negative feedback regulation certainly occurs and is vitally important. However, if negative feedback were the only mechanism that exists, regulation would be a sluggish and highly inefficient process. A physiologic insult great enough to trigger a reflexive response would have to already exist before regulation was initiated. The body would thus have to dedicate considerable resources (time and energy) to putting out fires, as it were. Luckily, that’s not how most regulation works.

Mechanisms of Physiological Regulation: Importance of Learning.

Converging research over the past few decades has revealed that it is preferable, and indeed more common, for organisms to take advantage of past experiences to enable them to prepare for potential challenges and ameliorate them before they occur; i.e., an under-appreciated mechanism for achieving physiological regulation is anticipatory responding, which allows for better compensation or preparedness for an impending physiologically-relevant stimulus (Dworkin, 1993). Learning on the basis of past experience enables anticipatory responses to be an effective strategy for achieving regulation because it allows effector responses to be activated without needing a deviation in critical physiological systems. Therefore, while many scientists continue to believe that regulation depends primarily on negative feedback mechanisms triggered by a deviation from a regulated variable’s set-point value, the bulk of evidence indicates that most regulation occurs without using negative feedback (Somjen, 1992). The reasons for the lingering general belief that negative feedback is the primary mechanism underlying regulation are discussed elsewhere (Carpenter, 2004; Ramsay and Woods, 2014; Siegel, 2008; Woods and Ramsay, 2007).

It is important to consider why it has taken so long for the scientific community to begin to embrace learning as a major regulatory strategy. For one thing, empirical learning as a scientific discipline was not known in Bernard’s time; and although Ivan Pavlov and Edward Thorndike, pioneers of empirical research on conditioning, predated Cannon’s writings, he seems not to have heeded their message. Another important point is that as learning became accepted as a valid process worthy of experimental investigation, mainly in psychology labs in the United States, it oftentimes seemed more like a parlor game. ‘Look, we can teach a pigeon to make a response when a light flashes; or a laboratory rat to …’ Psychologists felt a responsibility to use well-controlled and arbitrary stimuli that were neutral with regard to the task at hand lest the argument be made that they were stacking the deck in order to demonstrate learning. Light flashes, buzzers, electric shocks and many other stimuli not normally experienced by animals were commonly applied in laboratory studies of conditioning. The use of highly controlled stimuli, clearly defined and reliably measured dependent measures, and strong experimental designs provided convincing evidence of learning processes. Any semblance to the complexities encountered in ‘real-life’ situations was discouraged as that would result in less compelling (i.e., potentially more ambiguous) evidence of learning. One consequence was that an enormous literature developed on the properties of learning processes with little research attention or appreciation of the myriad ways animals actually use this learning capacity in ‘real-life.’

Acceptance of the importance of learning in normal physiological regulation started to change when it was found that organisms anticipating a meal secrete neurally-elicited insulin from the pancreas via the vagus nerves [see review in (Teff, 2011)]. This cephalic insulin, as it has come to be called, occurs just before actual eating begins and prior to any increase of blood glucose, and although not large in magnitude, its occurrence just at the start of a meal is sufficient to greatly limit the prandial glucose excursion. In fact, failure to make this simple anticipatory response renders a person or animal glucose intolerant for that meal, and to exhibit other symptoms of diabetes (Steffens, 1976; Lorentzen, 1987). Conversely, secreting insulin cephalically enables larger meals to be consumed without causing undue hyperglycemia, an adaptive response when food availability is uncertain (Woods, 1991). We (Woods et al., 1977) and others (see reviews in Teff, 2000, and Strubbe & Woods, 2004) have found that environmental stimuli that predict when food will become available, including the time of day, elicit cephalic insulin secretion at that time even if no meal is forthcoming. Importantly, in addition to insulin, many other hormones and enzymes that aid the digestion and absorption processes are also secreted cephalically in anticipation of meals, including ghrelin (Drazen et al., 2006), GLP-1 (Vahl et al., 2010), pancreatic polypeptide (Floyd, 1976; Teff, 2010) and glucagon (Fischer, 1976; Secchi, 1995), as well as meal-related neuropeptides such as hypothalamic NPY (Yoshihara et al., 1996 a; b). Cephalic secretion of so many meal-related metabolic compounds represents regulation at its best. When meals can be reliably anticipated, the consumption, digestion and further processing of the nutrients are less homeostatically disruptive. All of this is enabled by classical (also called Pavlovian) conditioning in which environmental stimuli become associated with the stimuli that activate already-established regulatory responses. In this example, time of day or the odor of food becomes associated with the hard-wired neural reflex by which the brain stimulates insulin secretion. Changes of nutrients or metabolic hormones directly in relevant brain areas, including the hypothalamus and brainstem, are presumably the genetically-determined stimuli that normally elicit insulin secretion via the vagus nerve, and through conditioning other stimuli (e.g., the time of day that food is regularly made available) acquire the capability to activate the same neural circuits.

Since the time of Pavlov, considerable research has been conducted to understand how classical conditioning occurs and to delineate the principles that underlie this type of learning (for a recent review, see Fanselow and Wassum, 2016). As discussed above, until relatively recently there was little recognition that classical conditioning plays a critical role in physiological regulation, and it was not until 1986 that Moore-Ede (1986) discussed the idea of predictive regulation in which responses can be activated in anticipation of a predictable challenge, heralding the importance of the role of learning in physiological regulation.

Ecological Relevance of the Conditioned Stimulus.

The relevance of learning to normal physiological regulation has been clarified by recent revelations on how conditioning actually occurs in real life. Domjan (2005) compellingly explained that in many laboratory-type conditioning experiments, tones, buzzers or lights, all of which are neutral and have no a priori link to whichever response is being assessed, are often used as stimuli. In contrast, animals in their natural ecological environment do not normally use conditioning to associate a once-neutral or arbitrary stimulus to a biologically meaningful regulatory reflex. Rather, animals co-opt environmental stimuli that have a natural relationship to the reflex. Thus, meal-related metabolic reflexes are readily conditioned to food odors or to the time of day that food is likely to be available; increased secretion of stress hormones readily becomes associated with the smell or presence of predators; and so on. The bias favoring ecologically relevant conditioned stimuli to form learned associations explains findings such as the ease with which odors or flavors, but not visual cues, can become associated with ingestible toxins in rodents (Garcia and Koelling, 1966). The point is that animals evolved to make conditioned associations between regulatory reflexes and ecologically relevant stimuli. While other more-arbitrary stimuli can in some cases become conditioned to elicit these reflexes, they are not the norm. As an example, when sounds (bell, metronome) were used in experiments by Pavlov demonstrating conditioned salivation in dogs, more than a hundred conditioning trials were often required (Pavlov, 1927). In contrast, neurally-elicited insulin secretion can become conditioned to odors within two trials (Woods et al., 1969). The concept of ecological stimulus relevance easily explains the discrepancy.

Generalization to Other Regulated Parameters.

In 1991, Woods recognized the functional similarity between the learned responses that prepare an individual to mitigate the physiological disturbance caused by eating with the learned responses that enable individuals to develop tolerance to drugs (Woods, 1991). In both of these situations, learned anticipatory responses defend physiologically regulated systems, thereby enabling individuals to tolerate the consumption of large meals or the administration of high drug doses. The importance of these anticipatory responses to successful regulation is made obvious when the predictive relationship between the two stimuli is experimentally manipulated. An unsignaled meal results in post-prandial hyperglycemia due to the lack of a cephalic insulin response (Teff, 2011). The presentation of meal-predictive stimuli results in cephalic insulin release, which if not followed by an anticipated influx of nutrients, can result in reactive hypoglycemia (Woods, 1983). Analogously, highly drug-tolerant individuals can defend critical regulated physiological functions while in the drugged-state due to making conditioned anticipatory responses (Siegel et al., 2000). However, these same drug-tolerant individuals, if they receive the drug without the drug-predictive cue, experience the effect of the drug unopposed by the learned activation of opposing responses, resulting in a deviation in a critical physiological system that is often misdescribed as drug-overdose (Siegel, et al., 1982; Siegel, et al., 2002). Conversely, when the drug-predictive stimuli occur but are not followed by the expected drug delivery, individuals can make anticipatory learned responses that are not opposed by the drug’s anticipated pharmacological effect, resulting in withdrawal-like symptoms due to the physiological disturbance caused by the individual’s own (in this case maladaptive) response (Lê, et al., 1979; Mansfield and Cunningham, 1980; Siegel et al., 2000).

Conclusion.

The fundamental importance of making learned anticipatory responses in regulatory physiology is becoming understood. Indeed, anticipatory responses are ubiquitous in regulation. However, recognizing anticipatory responses when they occur remains a challenge because the biomedical research field has been trained to view regulation from a negative feedback perspective. To broaden their perspective, investigators should ask themselves whether the change being measured in some parameter in an experiment represents the effects of a disturbance that elicits a corrective regulatory response (e.g., declining blood glucose levels that trigger eating) or whether it might reflect an anticipatory response that is preparing the animal for an impending disturbance (e.g., a drop in blood glucose caused by cephalic insulin released in anticipation of a meal). Experimental designs are available that can distinguish these alternative interpretations.