Nature is saturated with feedback mechanisms ranging in scale from the molecular to the macroscopic. Feedback is defined as the process of mutual causal interaction: where A affects B and B affects A. This interaction creates a circuit of effects, so any change in A, causes a change in B, which in turn causes a change in A.
Negative feedback loops, positive feedback loops, and interlocking loops on loops all together serve to maintain a system in a state of dynamic equilibrium. Acquisition of skills, brain development, cellular structure, ecological systems, and even human communication depends on feedback loops. For example, without feedback in the form of verbal and non-verbal cues from the listener, verbal communication would immediately slow down and stop. For this particular essay, I want to delve into the feedback loop particular to human reasoning.
Experiment after experiment over the last 50 years has shown that human reasoning is lazy and biased. Lazy because reason makes little effort to assess the quality of its justifications and the arguments it produces. Biased because it overwhelmingly finds justifications and arguments that support its own point of view. However, if human reasoning is so flawed, how could it have purported to elevate our species to the top of the ecosystem? After all, it is reason that sets us apart from other animals and is what makes us knowledgeable and wise. It is reason that facilitates better decision making. This view of flawed cognition flies in the face of evolutionary adaptation theory that states that traits (such as reason) are selected for by the “mechanism” of natural selection and provide fitness to the organism. In the Enigma of Reason, Hugo Mercier and Dan Sperber propose an interesting hypothesis that rethinks the function of reasoning. They state that the primary function of reasoning is not as traditionally thought to enable humans to go beyond mere perception, habit, and instinct but is primarily social in the form of devising arguments, persuading others of one’s rationalizations, and assessing the quality of other’s rationale. In effect, the process of back and forth arguments leads to a more coherent and sound hypotheses. Processes such as the Delphi Method that utilized this concept have shown to be more effective forecasters than solitary experts. Additionally, a recent case report showed the value of a shared discussion in developing more informed and sound decisions.
This has wide implications for a high throughput venue such as the emergency department. Unlike other venues in medicine, emergency physicians often do not have the luxury of time or resources to have a lengthy feedback process. In a given shift, emergency physicians make on average 10,000 decisions along along with 4000 clicks on their patient’s electronic health record. There is rarely, if any, time for feedback from nurses and patients and these emergency physicians are often the only doctors in the Emergency Department. These sole providers are often making time sensitive decisions based entirely on their expert intuition and subconscious biases. In the previous two essays, I wrote about cognitive biases, base rates, sample spacesand the utility of an intra EHR computational platform that can facilitate more stochastically accurate predictions by broad framing patient presentations as a tool for assisted decision making. Another feature of such a system would be an interactive feedback mechanism that would enable the physician to reason back and forth with such a platform. A platform that provides simulated predictions with physician controlled inputs could serve as a “pseudo-feedback” mechanism. Another feedback mechanism would involve warnings of potential decision pitfalls that would serve as an outsider evaluating the quality of the physician’s arguments.
The 19th philosopher David Hume noted that “truth springs from arguments between friends.” Arguments are feedback mechanisms that enable a group of humans to persuade each other of their beliefs and assess the quality of each other’s rationale and make a prediction on the future. It’s been said rather facetiously that “predictions are hard, especially about the future” but for solo forecasters this process is nearly impossible. Emergency physicians are without the luxury of “friends” in the emergency department and are often tasked with making complex decisions without the benefit of feedback mechanisms that enhance their solitary rationalizations. We form hypotheses on expert intuitions and biases and rationalize these hypotheses in the electronic health record but lack a realtime feedback mechanism that can evaluate the content of our argument. An intra-EHR computational platform that can provide cognitive support in the form of a “pseudo-feedback” could be a solution to limit the pitfalls associated with solitary decision making in a volatile, uncertain, complex, and ambiguous (VUCA) environment.