Skip to main content

Blog Post #3



This article discusses how, in the biomedical model of disease, instances of symptoms without a physical cause and relief through placebo intervention are anomalies. However, in the Bayesian approach to perception these two phenomena are not only explained, but accommodated for. The article states that in order to perceive the world, the brain follows a theory of probability known as Bayes’ rule. Having learnt about Bayes’ rule in class, the mathematical equation of it updates the likelihood of an event, given some evidence, by considering the product of the likelihood of that event and the previous probability of the event. Over rapid time scales, the brain uses Bayes rule to continuously generate a top-down cascade of neurally encoded hypotheses about the state of the body and the world. The top-down flow of hypotheses are met by a bottom-up stream of sensory inputs. Any mismatch between the predicted sensory input and actual sensory input creates a “prediction error” that prompts the brain to revise its hypotheses. Over the course of one’s lifetime, the brain continuously updates the probability of these events making it increasingly able to better predict the next sensory input and minimize error.

However, an implication of this theory is that what one perceives is not the world as it actually is, but instead the brain’s best guess of it. For instance, when faced with the task of determining the probability of an event given a set of inputs, the brain uses prior experience and contextual cues. The article highlights this through an example of an instance where the brain knows that a forest is full of snakes, and then sees sticks in the forest causing it to identify them as snakes. In this scenario the precise hypothesis shaped by prior experience and knowledge overrides the visual inputs.

This idea that what we perceive may not be exactly the world around us translates directly to the body and medicine. Bayes rule suggests that we feel pain because we predict that we are in pain, based on integration of sensory inputs, prior experience, and contextual cues. Thus the experience of being healthy depends on the fact we maintain a healthy bodily condition. As long as variations in bodily health remain within the acceptable range the brain treats fluxuations as irrelevant and no symptom is perceived. However, when a disease causes fluxuations out of the acceptable range of predicted health, the prediction error increases, and the brain updates its hypothesis. Following this reasoning, we feel symptoms only when the prediction error increases forcing the body to reevaluate its hypothesis. Thus, when a person unexpectedly experiences a painful sympton the hypothesis of health is quickly revised. For chronic symptoms slight variations in inputs, that would typically not be acknowledged, prompt the brain to mistakenly infer pain, and as a result revise the hypothesis to reduce prediction error, and nonconsciously create it. Similarly, with the “nocebo effect”, prior knowledge of the possibility of symptoms, for example known side effects of a medication, cause people taking  that medication to nonconsciously create them. Furthermore, from a Bayesian perspective the experience of recovery, in which the brain revises the hypothesis from sick to healthy again, occurs not from a direct return to health, but from the person inferring that they are improving. This revision is typically sped up when the person is giving external cues, such as a medication. A study shows that patients secretly administered symptom-relieving drug experienced much slower improvements than patients given the same drugs in plain view. Similarly, the brain believing it is improving may stop creating some of the pain it previously was creating, contributing to the improved health outcome.

In conclusion, Bayes rule applies not only to the mathematical reasoning under uncertainty that is the probability of an event, but also to both real and perceived diseases. In order to account for this approach however, medicine must consider the predictive process that lies at the basis of symptom perception, and evaluate courses of action that can lead the brain to predict the body’s health.


Leave a Reply

Blogging Calendar

November 2018
« Oct   Dec »