Chapter 1. Thinking Thoroughly and Methodically

Today’s culture increasingly alludes to algorithms—a set of rules or processes to be followed in calculations or problem solving. These underlie the computer programs and applications (“apps”) that we use daily but are really nothing more than equations and step-by-step instructions humans have always used to achieve goals. The process of becoming the finest clinical psychopharmacologist you can be will involve developing and internalizing your own algorithm—the most useful set of instructions to follow, every time, step-by-step, to produce the best outcome for patients.

Associated with algorithms is the concept of memes: systems of behavior that are passed from one person to another—not genetically but by imitation. The word was coined more than four decades ago (Dawkins 1976). Given our current internet and social media milieu, “meme” has itself become “viral” and has evolved to refer not to ideas but rather to a group of items created with an awareness of each other (Shifman 2013). In this chapter, meme refers to its original usage and definition in evolutionary biology. Our goal is to describe algorithmic changes you can make to your practice and carry forward as a meme that can be imitated and shared with the profession at large, improving the practice of psychopharmacology.

As we interact with the world of data, we seek to refine it into the world of knowledge. All data are not knowledge, just as all knowledge might not be wisdom. The clarification of knowledge from data, followed by its judicious use, is the source of wisdom. The most useful tool we have as clinicians in sorting knowledge from data is the modern scientific method.

Ancient philosophers explored the process of rational thought, seeking to develop reason so that human cognition could achieve a greater understanding of the world. Plato believed all knowledge could be obtained through pure reasoning. Aristotle influenced centuries of scholars by combining observation and measurement into inductive reasoning, realizing that rational thought had to be supported by data from the real world. Using inductive reasoning, an observation is made, and the mind attempts to draw a conclusion from the data observed. For example, every bald man you see might be geriatric. Therefore, inductive logic would lead to the theory that all bald men are of advanced age, a theory we know is not true when placed beside further observation or testing of this hypothesis. This represents a sampling error, which is solved with frequentism (discussed later).

Aristotle’s work was preserved and expanded upon by Islamic scholars such as Ibn al-Haytham, who formalized the first modern scientific method based on observation, hypothesis, and experimentation; Al-Biruni, who appreciated the opportunity for bias and stressed replication of results; and Ishaq bin Ali al-Rohawi, the first to utilize peer review. Progress continued with the Europeans Roger Bacon and Francis Bacon, who further used inductive reasoning in early Western forms of the scientific method. Renée Descartes is remembered for returning Platonic deductive reasoning to the Western method, which could then involve observable testing (Shuttleworth 2009b). Deductive reasoning is the use of hypothesis to predict observations, valid only if the hypothesis is true. For example, a hypothesis states that all As are Bs; if this is correct, then if C = A, then C = B. Deductive predictions can be logical yet false if the hypothetical premise is false (Bradford 2017).

Using various parts or forms of deductive and inductive reasoning, the scientific method was developed and improved, and knowledge advanced. Copernicus used his actual measurements and mathematical calculations to shift thought away from our egocentric cosmology and toward a heliocentric solar system. The tale of Galileo dropping his spheres from the tower of Pisa to show how experimentation triumphed over theory is well known, but he is also important for his standardization of measurements among experiments. Francesco Redi laid to rest the centuries-old debate over spontaneous generation by performing the first known controlled experiments. Isaac Newton formalized the synthesis of inductive and deductive reasoning into our modern scientific method (Shuttleworth 2009b). Louis Pasteur’s experiments led the world into a deeper understanding of microbiology, and Robert Koch’s postulates advanced the understanding of infectious disease transmission.

Physiology advanced through the work of rational scientists such as William Harvey (1628/1995), the first to describe completely how blood circulates through the body, who wrote, “I do not profess to learn and teach Anatomy from the axioms of Philosophers, but from Dissections and from the fabrick of Nature” (p. XIII), and Robert Hooke (1667), known not only for his description of the cell but also for his clever experiments testing Galen’s hypothesis about how the movement of air through the lungs was necessary to sustain life through its support of circulation.

Yet by the twentieth century, scientific philosophers such as Karl Popper had come to realize that science still was not infallible, that some disciplines advanced without using the scientific method and others that used it came to quite erroneous conclusions. He redefined scientific fields as those whose hypotheses could be tested through conjecture and refutation or “falsification.” These did not include psychology and social sciences until empirical methods were developed (Shuttleworth 2009a). Thomas Kuhn (1962) usefully explained how paradigm shifts described scientific revolutions: how stubbornly scientific communities hold onto current beliefs (strong priors in Bayesian logic, see Chapter 4) and the enormity of data required to dislodge them. Paul Feyerabend observed that not all scientific disciplines are able to follow the same rules. When it is very difficult to control all the variables necessary to form a clear test of a hypothesis, inductive reasoning must be relied upon more heavily in the balance—as Copernicus, Galileo, Charles Darwin, and Albert Einstein found when they developed hypotheses that took years to fully test (Shuttleworth 2009a).

In psychopharmacology, we are called to be rational and to use the same scientific method that led to the discovery of the Higgs boson and the structure of the atom. As we wait for our field to further improve empirical methods and to build enough data to solidify or tear down our current paradigms, however, patients are waiting for relief. We are obliged to offer them only the information that we have been able to test and should not rely on highly inductive guesses that will wait decades for verification.

As I discuss later in Chapter 4, we rely on probability theory, which was initiated in the seventeenth century by Blaise Pascal and Pierre de Fermat, advanced by Jacob Bernoulli and Pierre-Simon Laplace (names you would probably associate with other areas of scientific advancement), benefited from significant revision by Thomas Bayes and Ronald A. Fisher, and is still developing. Fisher is most associated with defeating the null hypothesis with P values and with advancing the frequentist approach that any outlying finding is likely to eventually be subsumed by larger and larger data sets. Unfortunately, not all knowledge can be obtained by restricting truth to a rigid cutoff for significance in order to defeat the null hypothesis: far too many theories must be tested, and our resources, especially for many possible randomized controlled trials (RCTs), are limited (see Chapter 2). Bayes’s prior probability (Chapter 4) can help rationally inform our use of the degree of significance in our abductive logical creation of treatment decisions. Abductive reasoning, explored further in Chapter 5, is what we often use when forming diagnoses and treatment plans: we develop theories from partial information, such as examination results and RCTs, and make temporary hypotheses that are then tested.

We must also take into account unconscious biases that may distort our interpretation of data and cause us to reject important information that we need in our knowledge base. The same may happen in our patients. Haidt (2012) wrote of how an emotional response, such as that linked to morality, will lead a person to cherry-pick data of varying quality to justify that response rather than evaluate the data objectively and seek more data to form or alter opinions. The use of reason, in this case, is in service of supporting the emotional reaction rather than the best objective knowledge. Furthermore, Schulz (2010) explored how social pressures shield individuals in groups from contrary data, exaggerating support for our own ideas and discouraging internal and external disagreement.

These influences are in addition to the cognitive confirmation bias that we attempt to eliminate in clinical research by, for example, blinding ourselves to the nature of samples during experiments. As I discuss further in Chapter 5, we strive to make treatment recommendations based on the results of controlled clinical trials, not on the hypotheses upon which they are based. Awareness of these emotional confounding factors in our assessment of data is the first step in minimizing their influence. Read broadly and discuss often with those who disagree with you on clinical issues, not so that you will automatically adapt your own thinking to theirs (they might be wrong as well) but so that you are exposed to the greatest amount of information, and attempt to consider it all value free (beyond the highest regard for the scientific method and good study design).

Although we might expect that clinicians’ knowledge base and wisdom only improve during medical education, data show otherwise. Misconceptions about the biomedical information to be absorbed and collated develop alongside more correct conceptions (Badenhorst et al. 2015). Furthermore, research about how we learn scientific information spotlights additional potential roadblocks to useful knowledge. It appears that learning new scientific theories does not result in the simple displacement of the original naive and, presumably, inaccurate theories but rather is added to them, requiring our brains to actively suppress the older theories for many years in order to recognize the new. The extended length of time this requires is measurable (Foisy et al. 2015; Shtulman and Valcarcel 2012). Even the best intentions must overcome many unconscious and potentially conscious barriers to understanding and using only the best knowledge.

It is unlikely that human clinical psychopharmacologists will be completely replaced by artificial intelligence (AI) in the foreseeable future, but tools to supplement our work involving machine learning and AI are already here. “Big data” also offers the promise of unexpected knowledge and speed of response. From the simple use of the internet to look up studies and dosages, to computer-assisted scans for drug–drug interactions, to electronic medical records and even email for clinical communication, algorithms developed by others are already sharing the workload with us. We must make sure that our own contributions are as reliable.

Our current situation really is not any different than when physicians looked up pharmacological data in hardbound reference books such as Drug Facts and Comparisons (Facts and Comparisons 2016), kept notes and prescription logs in paper charts, and stored paper copies of favorite journal articles in physical filing cabinets. The processes are similar; only the equipment has changed. A similar change occurred in the 1950s with the introduction of the copy machine and in the 1970s with the introduction of the answering machine. What we did and aspired to did not change, the processes just improved and made it a bit easier for practitioners to make decisions.

The president of Northeastern University, Joseph E. Aoun, Ph.D., has explored how higher education may adapt to preparing students to remain relevant as AI takes over more tasks previously performed by humans. He suggests that although knowledge alone will not be sufficient to reserve a place for the effort of the human mind, the ways we think about the world may remain our contribution. He suggests emphasizing not only entrepreneurship and cultural agility but also systems thinking or making integrated connections (Aoun 2017). He proposes that AI may analyze complex problems well, but it cannot yet understand how to apply the results in a creative and useful way through critical thinking. Although convergent thinking (identifying the single best answer) may sometimes be handled by AI, divergent thinking (generating multiple responses in a free flow of ideas) remains a skill humans might master. We can learn to understand and apply the context in which data are gathered and answers are formed, thus using well-analyzed ideas and information most effectively.

While we look at the cutting edge of the future with AI, algorithms, and memes, let us also carry forward the concept of a motto: a phrase guiding the ideals of an individual or institution. Visualize a scroll containing the Latin phrase penitus et methodice, omni tempore (thoroughly, methodically, all the time). Imagine this motto when you think of yourself practicing psychopharmacology.

The words thorough and methodical recur often throughout this book, and to that is added the imperative to perform to this level habitually. The best clinical practice involves repetition—repetition of all the known best practices, completely, every time. The clinical environment of excellence we are creating has no room for shortcuts—no hunches, no cutting corners “just this once.” As I cover in Chapter 2, we can never achieve certainty in science, and we can approximate it only with a rigorously applied scientific method. Probability will play a large role in the treatment recommendations we make to patients. Uncertainty and probability are inherent in our process, so we must leave no room for unnecessary expansion of uncertainty through incomplete assessments, inadequate knowledge of the literature, unrecognized bias, and the lack of a plan. The goal is to do things the right way the first time and every time to benefit our patients.

It can be useful to consider the best practices of successful practitioners in other fields. In a recent interview with Dinah Eng, the celebrated architect Frank Gehry discussed the importance of always being your professional best. He tackles each project, even small ones, the same way, as though it were the most important. He tries to avoid expectations in a project and tests relentlessly, not just following the first idea that comes to mind (Eng 2018). He is thorough and methodical.

To achieve this yourself, you must develop and refine your list of “best practices,” just as hospital quality and improvement committees do, and you must promise yourself that you will adhere to them, adopting them in toto as your algorithm and therefore demonstrating a meme—a system of behavior you can pass on. Identify your tools: trusted sources of information (your knowledge base), adequate resources for patient evaluation and monitoring (time, space, safety), and sufficient resources for treatment planning (trust, a good therapeutic alliance, clinical judgment). Patients bring to the therapeutic alliance their trust in us, their disclosure of information, and their agreement to comply with a mutually agreed-upon treatment plan. Clinicians bring not only their attention, goodwill, and common sense but also their tools: trust in their patients and the alliance, a database, safety, time, space, and clinical judgment guided by critical thinking.

By pledging to yourself that you will apply these tools thoroughly, methodically, and consistently, you create the boundaries and structure of your algorithm and enter the flow of information in medicine as a positive influence: the rational psychopharmacologist. The chapters that follow will fill in the details.

The best algorithm for the clinical practice of psychopharmacology is a defined set of thorough steps, followed completely each and every time, that may be passed on as a cultural meme, improving overall practice. The modern scientific method is our best tool for developing wisdom from data using a combination of inductive, deductive, and abductive reasoning: observation, hypothesis generation, and experimentation resulting in new observations that restart the process. Probability theory further informs the practice of psychopharmacology.

Clinicians must be ever alert for unconscious biases, both emotional and cognitive, that restrict access to information and rational conclusion. As technology takes over many of providers’ traditional tasks, divergent thinking may yet still lead to the creative and useful application of knowledge beyond the conceivable capabilities of AI. Habitually thorough and methodical, rational psychopharmacologists consciously define best practices as the elements of their algorithm: their database, space and time for safe and adequate evaluation, development of a therapeutic alliance, and application of clinical judgment guided by critical thinking.