Psychiatric training is generalist training.

First we’re physicians. This is an incredible starting point for anyone evaluating and treating individual humans for nearly anything. Many cognitive and emotional problems turn out to be medical. Psychiatrists are always conscious of medical issues, even when a patient sees us “just” for emotional or relationship issues.

Then there’s the four-year residency (at minimum, fellowship training is longer) to be a psychiatrist. Residency includes a wide range of clinical experiences: inpatient treatment of the severely impaired, emergency/crisis work, consulting on medical and surgical inpatients with psychiatric needs, psychopharmacology in outpatient settings, conducting several kinds of psychotherapy under faculty supervision, and some limited work with children. There are also experiences in community psychiatry, forensic/jail psychiatry, and more. The training is tailored to create well-rounded mental health generalists.

Are there specialists in each of these areas who know more, and have more experience, than a general psychiatrist? Absolutely. But no other mental health discipline enjoys this breadth of training.

Of course, not all psychiatrists remain generalists. Most focus on medication management in outpatient settings (although surveys indicate that many provide psychotherapy at least occasionally). Others work in hospitals, become full-time consultants, take administrative or academic positions, or limit their office practices exclusively to psychotherapy.

However, I’ll argue that being a psychiatric generalist is our highest calling, the best use of our unique training.

In 1966, Abraham Maslow wrote: “I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.”

That’s the drawback to being a specialist. We need specialists. But a specialist in psychopharmacology will look for medication solutions to problems. A specialist in psychotherapy will look for psychotherapy solutions. Alternatives are apt to be overlooked either way. Or worse, certain treatable conditions won’t be recognized at all if they fall outside one’s domain.

In contrast, a generalist has a variety of tools, not just a hammer.

I’m still a physician when I conduct psychotherapy. I can detect signs that a medical issue may be affecting how my patient is thinking and feeling. Many medical conditions can affect mood, anxiety level, sleep, and so forth. Or maybe the medical issue I notice is unrelated to what we’re talking about in psychotherapy, but it’s helpful that I notice it anyway.

Conversely, I’m still a psychotherapist when I manage a patient’s medication. I may detect emotional conflicts that interfere with taking medication reliably. Or I may perceive interpersonal or self-esteem issues unrelated to what I’m treating pharmacologically. I may raise these issues with the patient as any psychotherapist might, or we could even convert the medication visits to psychotherapy visits if indicated.

This isn’t dilettantism. It’s not dabbling, trying a little of this and a little of that. On the contrary, it’s wisely choosing tools out of one’s toolbox. It needn’t be a single tool: I can do medication management and psychotherapy with the same person. But I’m not throwing everything at the wall to see what sticks.

It’s important to note how one’s tool choice affects the patient. For example, it would likely confuse and discomfort a patient if their psychiatrist suddenly and unilaterally shifted from being a chatty, casual medication-management doctor to a stone-faced, reserved psychotherapist. I’ve similarly heard of psychotherapists who decided to insert cognitive behavioral therapy (CBT) into one psychodynamic psychotherapy hour “to see how it would work.” Abruptly switching gears is confusing for patients and often counterproductive. It’s fine to introduce new ideas and try different approaches in a psychotherapy. But this requires sensitivity and is not casual.

Many years ago, an online critic derided being a psychiatric generalist as akin to a spork, a combination spoon and fork that doesn’t do either job well. I countered that a better model is a Swiss army knife, or a toolbox. Not one tool that does several things badly, but a collection of tools each honed for its specific use.

Since none of us has a toolbox (or Swiss army knife) big enough to hold all potentially useful tools, we must not denigrate our fellow mental health professionals. It’s important to know and respect what others can provide. I occasionally conduct couples therapy but often refer couples to colleagues. I routinely refer out patients who need a treatment I don’t provide. I refer patients for CBT when this seems in their best interest. I refer anyone who needs ketamine or transcranial magnetic stimulation (TMS) for depression. I don’t offer those, so it’s important for me to know and respect clinicians who do.

Some argue that as rare and expensive mental health clinicians, psychiatrists should limit our practices to “medical” psychiatry, usually taken to mean psychiatric medication and similar somatic treatments, and/or that we should oversee other mental health professionals as administrators or medical directors. In particular, they argue that psychotherapy is available from a large number of less expensive professionals. They say we should avoid doing psychotherapy in order to work “at the top of our license.”

These critics fail to appreciate that their argument applies equally well to medication management. Primary care physicians and nurse practitioners already prescribe far more psychiatric medication than we psychiatrists do. The point is not that we have exclusive skills—we don’t. Our superpower is breadth. That’s the “top of our license.”

When I was in medical school in the 1980s, I learned that about 85% of all problems brought to a general practitioner or family doctor could be handled by that doctor. Specialists were for the remaining 15%: diagnostic mysteries, unusually complex cases, or those who needed surgery or other specialized interventions.

The advantages of the old-fashioned medical generalist were obvious. There’s great value in a doctor who knows patients over time and sees the big picture, who can put a medical complaint or problem into context. It’s uniquely helpful for a doctor to coordinate the management of multiple medical problems so they don’t clash, and to provide sophisticated triage to specialists only when needed.

But as we all know, that ship has sailed. Primary care doctors are usually rushed, running behind schedule, distracted by the voracious electronic medical record, and struggling under caseloads so large they can’t really know their patients. Many verge on burnout. Meanwhile, the public has suffered to the point that many would sooner skip the gatekeeper and make an appointment directly with a medical specialist after consulting Dr. Google, or now Dr. ChatGPT. The exception to all of this is membership in a concierge practice. Few can afford it.

Mental health care is careening down the same path. Many psychiatrists are essentially concierge doctors, opting out of insurance panels and accepting only cash-pay patients. We do this for the same reason some of our primary care colleagues do. It’s the only way within our dysfunctional health care system to work as we were trained: as flexible generalists, not cogs in a corporate wheel. We choose not to be rushed, behind schedule, distracted, struggling, and burned out. But we recognize that few can afford us.

I respect my psychiatric colleagues who compromise their medical ideals in deference to their social justice ideals. But I identify more with those who hold that it’s not our primary job to fix society—that medical ethics prioritizes our obligation to the individual patient.

This view argues for the psychiatric generalist: the mental health professional who wields the widest vision on behalf of the individual patient. Admittedly, it’s not the most economical or time-efficient way to practice: Henry Ford invented the assembly line for a reason. But it is the best way, I believe, to treat an individual.

Yes, we need specialists for that last 15%. Meanwhile, we’re the clinicians, in theory at least, for the other 85%. We’re mental health generalists, with a purview that spans the molecular, to the patient’s inner emotional world, to the social environment that affects each of us in different ways. If only the health care system, and society at large, knew what they were trading away.

Recently a psychotherapist on LinkedIn wrote that after working online most of her (short) career, she has once again started seeing patients in person. Apparently she’d been typing verbatim notes, i.e., a transcript, while conducting sessions over video. Now she’s struggling to handwrite the same thing in front of patients in her office. She asked her fellow therapists: “How do you manage writing notes in face-to-face sessions without breaking your fingers?”

Here is my answer, along with some other thoughts about psychotherapy notes.

Aside from initial visits, I’ve never in 35 years taken notes during a psychotherapy session. Instead, I handwrite a short note immediately afterwards. That way I can devote full attention to the patient in the moment. Writing the note just after the session is important. Psychotherapy sessions are like dreams: they’re not hard to recall immediately afterward, but they “evaporate” quickly. When I’ve been delayed in writing a note, I’ve been able to do it, but not as well.

Even a fast typist can’t transcribe a psychotherapy session in real time—not while being present in the session as a therapist. Audio or video recording is much better for that, although it’s rarely needed outside of research settings or certain styles of psychotherapy supervision.

There are also “scribe” apps, many now with AI, that not only record but also organize and summarize therapy transcripts. I’ve never used one, and see no need. As noted in my previous post, it’s important that such apps preserve patient privacy, e.g., are HIPAA-compliant and do not share therapy recordings or their derivatives to train AI. Such apps also spare the clinician the cognitive work of organizing his or her thoughts about the session. This is a mixed blessing at best: it saves time and effort at the cost of engaging less deeply with the material. Also, the AI may not “conceptualize” the material the same way the clinician would, which in theory may affect the way treatment progresses.

In my experience, taking brief notes (about 7-10 lines... five minutes) immediately after the session is enough. When I review them before the next appointment, they jog additional memories, especially if I was paying full attention in the first place. This is also what I advise my trainees, and it seems to work for them too. Excessive note-taking is unnecessary and often bespeaks anxiety in newer psychotherapists.

In an institutional setting, or in any agency or group practice really, there are important benefits to keeping two sets of notes: the chart (or file), and separate process notes (or psychotherapy notes). The former is the official treatment record. It legally documents the professional service, and is shared with other “covered entities” under HIPAA. Typical guidelines are to document the facts: the session date, time, and duration, the nature of the service rendered, clinically pertinent facts such as medical issues and medication changes, risks such as suicidal or homicidal ideation, a brief assessment statement, and the immediate plan (return in one week, refer to Alcoholics Anonymous, etc).

Process notes, in contrast, are working notes written by and for the psychotherapist. They are not part of the official treatment record, and are not released to other clinicians for coordination of care, nor by patient request (however, they are subject to discovery by court order). They are the clinician “thinking out loud” about the case. As the name suggests, these notes record the process of the session: the topics discussed and the transitions from topic to topic, pertinent behaviors—the patient arrived late, fell silent after an outburst, was more jovial than usual, seemed guarded, forgot their checkbook, was unusually engaged, asked for advice, etc etc.—and even speculations, reveries, and countertransference feelings of the therapist.

Separating chart notes and process notes excludes sensitive material from the official record, yet the clinician is free to privately note speculative and highly sensitive material in order to personally reflect on the case.

But I confess that I’ve never kept separate process notes. As a solo practitioner, I’m the sole viewer of my notes 99% of the time. So my notes are a combination: a few lines that encompass the “facts” as well as some of the process. I omit particularly sensitive material (detailed dream reports, sexual matters) that I might have included in separate process notes. I tend to remember these details anyway. And in the back of my mind, I remain aware that the patient may someday request a copy of these notes, which they have every right to do.

However, I worry a bit about that remaining 1%: when a patient requests their file, or wants it released to a third party. I offer to write a one-page summary instead, and sometimes that’s enough. When it’s not, and the patient receives a complete copy of their chart, I offer to go over it with them, even if they are no longer my patient.

I’ve also received subpoenas for records several times over the years, i.e., very rarely. Generally this happens when my current or former patient is involved in a lawsuit, either as plaintiff or defendant. I invite the patient to have their lawyer attempt to quash the subpoena, assuming it isn’t theirs. Barring that, I copy and release the file, always informing the patient first. This has happened two or three times in my career.

Process notes are particularly useful for formal psychotherapy supervision. A colleague of mine long ago published1 on the unique pros and cons of process notes, audio recordings, and videotape in supervision. While hearing and seeing the “play by play” of the session clarifies transient dynamics, there is never time to view or listen to a whole therapy hour in supervision. Audio and video recording, and by extension verbatim transcripts derived from them, are best for presenting short excerpts, say the first five or ten minutes of the hour. Only process notes convey the flow of the whole session, as well as the unspoken thoughts and feelings of the therapist. Process notes also assist trainees in creating written psychodynamic formulations. This is an essential part of training, and often valuable in practice as well.

While documentation is essential for clinical, legal, and training purposes, it should never take precedence over the actual clinical encounter. The stereotypical “therapist with a clipboard,” as shown in the stock photo included here, is less than optimal in my opinion. It’s far more important that the psychotherapist be present, attentive, and attuned in the moment, engaged with the patient, than to be preoccupied collecting data for the chart. Even splitting attention between the patient and the clipboard is an unneeded trade-off. It’s a pity if the record of a basically sound treatment proves incomplete or unhelpful; it’s a tragedy if a failed treatment ends up perfectly documented.

I join most mental health clinicians in decrying “AI therapy.” There are many reasons why. General-purpose chatbots, ChatGPT and the like, lack confidentiality. Tuned to be agreeable, even sycophantic, they can reinforce the inclinations of suicidal or delusional users. They sometimes “hallucinate” and respond in weird, counterproductive ways that could confuse or agitate an emotionally vulnerable person.

I agree with these concerns, and I also highlight what I believe is an overlooked distinction: that chatbots like ChatGPT or Claude can provide emotional support but not psychotherapy. They can’t conduct a comprehensive psychological assessment; they have no strategic treatment plan; they can’t gauge progress or change. Much of the debate about their use conflates support and therapy. See my post “AI therapy” is Not Therapy.

Meanwhile, some therapists quietly defend AI as helpful—not as a therapist itself, but as their assistant. They rely on AI for “curbside consults” or psychotherapy supervision. They borrow clinical ideas or suggestions from AI. They use AI “scribes” to document and summarize sessions. And they use AI for clerical tasks. So in this piece, I’ll sideline the issue of “AI therapy” and discuss these other uses.

General concerns regarding AI

Some criticism of AI is not about its output, but about the resources it uses. Artificial intelligence consumes vast amounts of electricity for computation, and water to cool its circuitry. Fundamental alterations of national electrical grids are underway just for AI. Energy and water costs are rising in some areas as a result. Some say the worst thing about AI is its environmental impact.

However, it’s worth noting that resource consumption for text output pales in comparison to resource use in creating still images, which in turn pales in comparison to making AI-generated videos. According to one estimate, a short video uses 2000 times the energy of a text response. While a purist would avoid all of these uses, a more moderate approach would shun the widespread, frivolous creation of images and videos, and reserve this powerful, energy-hungry technology for legitimate social goods. Medical and mental health care would clearly qualify.

As another concern, critics warn of loss of human skills if we delegate them to AI. If AI formulates our clinical cases, if it writes our notes, if it does all of our thinking for us, these skills may wither from disuse. There’s well-founded concern that high school and college students no longer learn how to research and write term papers; instead they learn how to prompt AI to do it for them. Will psychotherapists in training likewise learn how to prompt AI, not how to think like a clinician?

And more: Every minute in front of a screen is a minute not spent interacting with other people. Concerns about the impact of screen time on our culture and on our relationships go well beyond AI. But AI clearly magnifies the problem.

Readers may decide on the basis of such issues to shun AI completely. It’s a consistent stance to take, just as vegetarianism is a consistent response to animal suffering. But like vegetarianism, avoiding AI completely is a signifiant limitation that most won’t endorse. As I describe below, it has become quite hard to avoid AI. Perhaps you’ll accept the above concerns as “the cost of doing business.” Maybe you just don’t care. Or maybe you’ll use AI sparingly, aiming to mitigate these drawbacks while still using AI in beneficial ways.

AI as mental health “consultant”

If you needed to know who proposed the idea of the “corrective emotional experience,” as I wondered recently, where would you even start to look? The answer, of course, is that you’d google it.

Web searches are so commonplace that google is now a generic verb. But search has changed. Google and Bing, the two most popular search engines, now present an AI-generated answer at the top of every results page with no obvious way to turn it off (I admittedly didn’t research this deeply). DuckDuckGo, a Google alternative that doesn’t track users, also provides an AI-generated answer by default, although it can be switched off. Ecosia, an environmentally-friendly search engine, has AI off by default, but you can ask for it.

AI is so ubiquitous that it’s hard not to use it for this type of information. These tools present and summarize answers faster than “manually” compiling them from search results. I usually turn to the free version of Perplexity for these questions, as it provides links to the references it used to create its answer. I sometimes check those references to confirm the quality and believability of its response. As we all know, AI sometimes presents completely made-up information in a persuasive way.

Artificial intelligence is hardly limited to search engines. It now embellishes messaging and office productivity apps such as spreadsheets and word processors, teleconferencing apps (e.g., Zoom), web browsers, computer operating systems, and so on. I consider this sea of AI excessive and wasteful, and rarely use these add-on features. Yet they’re there, eating up energy and computational resources. And they’re quickly becoming the norm and harder to sidestep.

Clinical consultation is a whole different matter. General-purpose AI chatbots (ChatGPT, Claude, Gemini, Perplexity, and others) are not HIPAA-compliant. They do not treat transcripts as private. Companies store and review user input and AI output for a variety of purposes, including to sell to third parties or to train the next generation of AI. Sharing protected health information (PHI) with ChatGPT in order to obtain consultation or supervision is a clear HIPAA privacy violation.

Since these chatbots are as close as your phone, tablet, or laptop, it’s easy to forget that the real work happens in a faraway data center. Using one feels private, but it definitely isn’t. Chatbots are arguably the adult equivalent of an attractive nuisance, the legal term for a hazardous object or condition that attracts children who are unable to appreciate its risks. (Classic examples are an unfenced swimming pool, or dangerous playground equipment in a public park.) Under tort law, owners of such hazards can be sued for the injuries and deaths they cause.

Is there a safe way to use AI for consultation regarding a real case? Yes, if you use HIPAA-compliant AI. Both Open Evidence and Doximity offer this for health care personnel. These models train on reputable medical literature, not Reddit and Wikipedia as ChatGPT does. They treat PHI just as securely as other HIPAA “covered entities,” such as insurance companies and pharmacies (for better or worse). Plus, they’re free.

Although geared toward medical care, both Open Evidence and Doximity accept psychological case data and can provide a case formulation from a user-specified theoretical perspective (if that perspective is described in the medical literature). I briefly tested both systems and can confirm they work for this purpose, although I can’t vouch for their consistent accuracy or usefulness. Of course, they are only as good as the material fed into them. Present your case differently, and you are apt to receive a different response from the AI.

Supervision of a sort is also possible, although the hazards and shortcomings multiply. The AI accepts the user’s assessment of the encounter; there are not “two sets of eyes” on the case. The AI is unlikely to identify or correct a technical lapse by the therapist unless you note it yourself. At this point in its development, AI offers, at best, very superficial psychotherapy supervision. In addition, parroting AI-generated lines in the next therapy session will make the therapist sound wooden and unnatural. Suggestions from AI, or even from human supervisors, should be reworked to make them genuinely one’s own.

AI scribes

Another popular use of AI is as a scribe to document and summarize treatment encounters. Scribes arose as an outgrowth of electronic health records. Physicians were so burdened by documentation requirements that some hired human scribes to join them in the exam room. The scribe recorded the session and dealt with the EHR, freeing the doctor to interact with the patient.

Not long after, automatic recording systems in exam rooms replaced many human scribes. And now with the incorporation of AI, these systems systematically organize and summarize the recordings for the EHR, and also sometimes for patient handouts and the like.

Some mental health professionals join their medical colleagues in relying on these systems. I personally still use paper charts, and rarely give a patient anything written to take home. But therapists who treat patients who benefit from receiving a brief written summary after a session reportedly find AI scribes helpful.

Again, the key question is whether such a system is HIPAA-compliant. Legitimate AI scribe apps for clinical offices always are, but it’s important to confirm this.

Clerical assistance

As in other office settings, AI has proven valuable in mental health clinics and private practices for scheduling, client appointment reminders, billing, and the like. HIPAA compliance is crucial here as well, as many AI-based office management systems are not designed for health care specifically.

Between-session support

Rather than replacing psychotherapists, some AI apps offer between-session emotional support, psychoeducation, and/or symptom check-ins. Depending on one’s caseload and style of practice, these apps may enhance patient care. Refer to reputable reviews to find professional-grade products that suit your practice and the clinical needs of your patients.

Final thoughts

Once we dismiss the fantasy of psychotherapy delivered by ChatGPT, mental health professionals are left to weigh the degree to which AI will influence our practices. A few of us will manage, with careful and sustained effort, to avoid AI completely. At the other extreme, a few will throw caution (and patient confidentiality) to the wind and casually use whatever AI is at hand. The rest of us have hard decisions to make.

Most patients already use AI for work, for recreation, and some for companionship or emotional support. It helps to be familiar with this popular technology, just as it helps to be aware of our patients’ interpersonal milieu. It may also be clinically meaningful to explore their experience with AI: why they use it, how it makes them feel—the pros and cons from their perspective.

If psychotherapists contemplate AI with “evenly hovering attention,” neither with alarm nor excessive enthusiasm, we’ll empathize more deeply with our patients, and better understand the role of this technology in their lives.

Psychiatry at the Margins by Awais Aftab MD is a must-read for those interested in the philosophical underpinnings of psychiatry. His latest post recalled an idea of mine from long ago.

The year was 1989, more or less. Having completed my psychiatric residency at UC San Francisco, I continued there for two additional years as a post-doctoral fellow in psychotherapy research.

I joined a large multiyear project, led by Mardi Horowitz MD and funded by the MacArthur Foundation, called the Program on Conscious and Unconscious Mental Processes (PCUMP). The group micro-analyzed videotaped psychotherapy sessions, measured the in-session physiology of both parties, created meticulous transcripts and applied various discourse measures to them, and so forth. Horowitz envisioned synchronizing and combining all of this into an “orchestra score,” in order to find “flurries” of activity in the data that would point to meaningful moments in treatment.

Horowitz, a psychoanalyst and author of some renown, sought empirical support for psychoanalytic ideas, particularly the stress response syndromes and pathological grief reactions he had written about. He referred frequently to states of mind: the sum total of thoughts, feelings, wishes, associations, etc. in a person’s mind at a given moment. He was particularly interested in mental state transitions, the flow from one mental state to the next. It’s a notion that has figured prominently in my thinking ever since, particularly the irony that psychodynamics pays scant attention to the fine-grained dynamics of the psyche. Unfortunately, despite a number of papers and other publications from PCUMP, I’m not sure Horowitz ever found the Rosetta Stone he sought.

Others of us had our own agendas. Our second-in-command was deep into the computer technology of the day, and outfitted us with fancy computer workstations. I learned some Unix: grep, awk, regular expressions, shell scripting. Our statistician taught us about inter-rater reliability and analysis of variance. Our psychophysiologist had previously been the 13th employee at Apple Computer, had studied altered states of consciousness with Professor Charlie Tart, and occasionally took off on side projects, for example to record the physiology of advanced yogis in India.

It was a smart, fun group, and I thrived in the research environment. I loved crunching numbers and thinking big thoughts about our field. I even enjoyed staff meetings.

As a post-doc, I was assigned a faculty advisor to guide me in developing my own research project. I proposed the idea that psychiatric diagnoses are prototypic. My thinking went like this:

A diagnosis, such as major depression, is a conceptual prototype, an exemplar. A patient’s collection of signs, symptoms, and states of mind approach this prototype to a greater or lesser extent. I pictured these exemplars as mountain peaks in a mental landscape. A person’s psyche could be in a valley, not near any diagnostic peak; partway up a slope toward a diagnosis; or close to the top, nearly a classic case. Two or more diagnoses might sit close enough together in this landscape that their mountain bases overlapped somewhat, such that a given patient could be partly up more than one. However, paradigmatic cases would always approach the top of a single peak.

The model was inherently dimensional. Resembling an exemplar—goodness of fit—is a matter of degree. At some arbitrary height up the mountain, the person’s condition is deemed major depression; at lower elevations they’d be mildly depressed or even normal.

This idea partly followed from DSM-III, the third edition of the Diagnostic and Statistical Manual of Mental Disorders, published in 1980. This was the edition that introduced diagnostic criteria. If the patient fulfilled five of the possible nine listed criteria for major depressive disorder, they “met criteria” for that disorder. My model of prototypicality was similar, but more nuanced I believed. There didn’t have to be a specific number of criteria, they could differ in importance, and some features could be interpersonal or culturally-bound; for example, evoking a particular flavor of countertransference.

In any case, my faculty advisor didn’t know what the hell I was talking about. He suggested I pursue something more concrete. I let the prototypicality thing drop, but mustered no enthusiasm for anything simple or concrete.

I eventually found my niche after reading James Gleick’s 1987 bestseller Chaos: Making a New Science. I realized that moment-to-moment mental states transitions (a la Horowitz) might be analyzed using nonlinear dynamical modeling, i.e., chaos theory. Dr. Dana Redington, our psychophysiologist/Apple programmer/fellow big thinker, joined me in this. Using nonlinear mathematical modeling that was rather beyond me, he analyzed parasympathetic nervous system tone, derived from heart rate data he’d already collected for PCUMP. We then related “trajectories in the psychophysiological phase space” to the clinical material in the videotapes.

We were a bit like Wozniak and Jobs at Apple. Dana did the tech, I presented what I hoped were the exciting clinical implications. I wrote about how point attractors in this phase space corresponded with oft-visited states of mind, how different clinical states (avoidance, engagement) observed within a psychotherapy session had phase-space correlates, and how the total phase space of mental states visited by a person during their life constitutes that person’s personality.

Eventually we published three peer-reviewed papers, a book chapter, and a few other papers on this. It was cutting-edge stuff at the time: we made the cover of Science News in 1991. But, alas, our lives and interests diverged. Dana became a computer consultant, I pursued clinical psychiatry. We did no further research. As it happens, a few years later we both embarked on lifelong studies of Aikido (in different dojos), but that’s a story for another day.

I haven’t thought much about diagnostic prototypicality in the years since. Thanks largely to Psychiatry at the Margins, I realize that even prototypes or exemplars don’t solve many of the problems of psychiatric nosology. Writing about it now, I’m struck by the parallel between diagnosis as prototypicality—mountain peaks in a multidimensional landscape of signs, symptoms, and states of mind—and nonlinear modeling of mental state transitions in an individual, a complex trajectory flowing through a similar multidimensional landscape.

I haven’t conducted research in over 30 years. Others continue to apply nonlinear dynamical modeling in the physical and social sciences, including in psychology, although I’m not aware that anyone has pursued exactly the direction we did. A more useful explanatory model involving prototypes or phase-space analysis may still be out there, waiting to clarify our understanding of mental states, personality, and psychiatric disorders. If we can revisit the moon after 50 years, why not moment-to-moment mental state transitions after 30?

On March 30th, the New York Times published a guest essay (formerly known as an op-ed) by Divya Saini and Natasha Bailen highlighting the drawbacks of using AI for emotional support. Their essay begins with seeming advantages. Chatbots are always available, “inhumanly” patient and validating. Unlike humans, chatbots never tire of hearing the same complaint. They never get frustrated or angry. In short, “it can be easier to turn to a computer rather than a person.”

On the other hand, chatbots can treat delusions “as a plausible premise to explore rather than a flawed perspective to gently challenge.” They can strengthen delusions and push vulnerable users toward suicide. They may fail to refer users in crisis to mental health care when it is indicated.

More generally, the authors caution that turning to chatbots for emotional support results “in patterns of reassurance-seeking and rumination that are hard for people to recognize in themselves.” They cite recent research showing that prolonged use is associated with increased emotional dependence, social isolation, and loneliness.

These are all good points. More could be said about data privacy and other concerns, but the essay is fine as far as it goes: AI chatbots offer helpful emotional support for many users, while introducing risks that we should take seriously and attempt to mitigate.

Unfortunately, the Times editors confusingly titled the piece “Your Chatbot Isn’t a Therapist.”

The authors never once mention chatbots posing as therapists, nor users treating them as therapists. The word “therapist” appears nowhere in the essay, and “therapy” appears only once, in reference to actual human therapists.

Nevertheless, many reader comments that follow the essay refer to AI-delivered therapy, perhaps primed by the misleading title. A clear majority defend it, citing the scarcity, cost, and sometimes the incompetence of human therapists.

In other words, readers of the Times are poised, with very little prompting, to extol the virtues of therapy delivered by AI chatbot—despite the fact that general-purpose chatbots deliver no such thing. This reflects a fundamental confusion about what psychotherapy is, and how it differs from emotional support.

Emotional Support

Emotional support is, or should be, widely available in everyday life. We are social animals by nature. We seek and receive emotional support from friends, family, romantic partners, religious and spiritual communities, social clubs, political allies, coworkers, bartenders, hairdressers, and from society at large. In addition, our natural environment, our pets, our parks, our beaches and gardens support us emotionally. And humans support each other vicariously through art, literature, music, and drama.

For a number of decades, observers have lamented the erosion of community—that is, emotional support—in our society. Families are fragmented, organized religion is in long decline. As bowling leagues and bridge clubs die off, more of us live in small nuclear families or alone, entertained by television and the internet. We’re physically disconnected, our social lives delivered by proxy through screens.

Chatbots partly fill this dearth of emotional support. That’s largely why they’re so popular. Unlike television or much internet content, chatbots are interactive in real time. They emulate human conversation, capitalizing on our innate gift of imagination and our ability to anthropomorphize. Just as we do with our pets, cars, Siri, and Alexa, we pretend that we’re talking to another person. It’s sort of fun. But AI designers go much further. They tune chatbots to be agreeable and pleasant, even sycophantic. Some chatbots can speak their responses using vocal tone and prosodics that mimic the real thing. All of this keeps users more engaged, the ultimate aim of AI developers.

At one level, it’s sad that the lack of real human connection leads so many of us to turn to mechanical simulacra for emotional support. It’s a band-aid covering a deep wound. But AI does fill the need, partly and imperfectly. Many people find chatbots supportive and helpful despite the potential for danger and harm.

How Did Psychotherapy Become Equated with Emotional Support?

Psychotherapy and emotional support are very different things. However, in recent years the distinction has eroded in the public eye. Several factors account for this, many from the profession itself.

Humanistic and client-centered approaches to psychotherapy, entirely legitimate as originally proposed, paved the way for thoughtless loosening or abandoning of technique. Some psychotherapists came to believe that in order to promote healing, nothing more is needed beyond being present and “empathic,” i.e., emotionally supportive.

Self-described “trauma-informed” psychotherapists1 avoid pathologizing clients, which they mistakenly equate with “blaming the victim.” In claiming that all woes are normal reactions to abnormal situations, there is nothing to treat in the client. The role of the clinician is reduced to validating and normalizing the client’s feelings of victimization. In other words, offering emotional support.

Outside the consulting room, psychotherapists learned to self-promote. These days we all have websites and online directory listings for our practices. Some of us are on social media. Now that we advertise in various ways, emotional support is a lot easier to sell than the hard work of psychotherapy. We compete against large corporations such as BetterHelp, which touts emotional support almost to the exclusion of psychotherapy. They know what sells.

Last but not least, influencers on social media reduce psychotherapy to “therapy.” Although this sounds like simple shorthand for psychotherapy, “therapy” covers much more. It’s a loose catchall for coaching, advice, inspirational tidbits, cognitive exercises, casual and oversimplified use of clinical terms such as “narcissism” and “psychopathy”, and so on. I call this “therapy that isn’t psychotherapy.” It’s basically emotional support.

So what is psychotherapy? It’s a clinical service, delivered by a trained health care professional, based in a human relationship with the patient or client. It’s a quasi-medical treatment for psychological disorders. Psychotherapy has clinical aims above and beyond helping the patient or client feel good in the moment. It aims for lasting change.

Because psychotherapy is a treatment, by definition it treats something that is wrong in the client or patient. Patients usually recognize this themselves. Fear of pathologizing (“blaming the victim”) sidesteps this reality and fails to distinguish what we do from emotional support offered by family, friends, or a sycophantic chatbot.

It makes no sense economically to seek emotional support from an expensive mental health professional. Such support is available elsewhere at far less or no cost, and this isn’t a psychotherapist’s job in any case. We are supportive at times, but only as a component of treatment. Instead, we offer far more: release from self-defeating patterns, easing of dysfunctional views of self and others, increased productivity and satisfaction in life, resolution of anxiety-provoking inner conflicts, and so on.

“AI Therapy”

When an opinion piece on emotional support from chatbots leads to dozens of public comments praising nonexistent “AI therapy,” it’s clear that much of the public doesn’t know what psychotherapy is. This isn’t new: people have long joked that their bartender or hairdresser is a better therapist than their actual therapist. Psychotherapy is often mistakenly equated with emotional support. Judged by that standard, it will often be true that bartenders and chatbots do a better job.

Mental health professionals and our professional organizations need to educate the public that emotional support, and the sycophantic support of AI chatbots in particular, isn’t therapy. This isn’t an easy task when some human therapists appear not to appreciate the distinction in their own practices; when much of the public recognizes no distinction; and when powerful social media messages preach exactly the opposite. Yet if we do nothing, this confusion will further degrade our profession.

We cannot compete with chatbots, any more than we can compete with family, friends, or bartenders. Not because they perform our work better than we do, but because they don’t perform it at all. We have a lot of work ahead of us, to make the case that we provide something more and different than mere emotional support. We should probably start with our own colleagues who vainly attempt to compete with chatbots and bartenders.

Pop psychology, the lifeblood of social media, repeatedly reinvents the wheel. Influencers hype the latest buzzword as a new discovery to revolutionize our understanding of ourselves and others. In order to do this, they coin a provocative new name and couch their invention in popular, easy-to-digest concepts. Functional freeze is a paradigm example.

“Functional freeze” is the novel term for functioning normally—or nearly so, as I’ll discuss below—but feeling emotionally numb, disconnected, or stuck internally. It’s a feeling of paralysis in decision-making and motivation despite appearing productive, often accompanied by vague anxiety or an ill-at-ease feeling. The sufferer is still functional while feeling frozen inside.

According to proponents, functional freeze is a stress response. Framing it this way fits popular notions regarding trauma, burnout, and the like. That is, it reflects the prevailing pop-psychology view that people mostly suffer in response to external stressors, and that it’s our job as therapists, or as the folks who sort of act like us on social media, to validate and support those sufferers. I criticize this insipid view of psychotherapy on a regular basis.

The argument begins with science, then veers off-base—first a little, then a lot. Acute stress responses, mediated by the sympathetic nervous system, were once described simply as “fight or flight.” These remain the two most common responses to a sudden adrenalin dump: in people, becoming enraged and combative, or running like hell.

But in recent years more possibilities have been added. Under acute stress, animals and people can also freeze: picture a deer caught in the headlights, or someone too scared to move. Another possibility is fawning: trying to please or appease the threat to avoid conflict or harm. And some propose adding options such as flopping (fainting), flagging (becoming discouraged), and even fine, which refers to being in denial about the stress afterwards.

It’s a tribute to alliteration, if nothing else.

This is classic concept creep. The fight or flight response is a physiologic feature of all animals from flatworms and sea anemones on up. Animals with more complex nervous systems also sometimes freeze. But by the time we get to fawning or flagging or “fine,” we’re really talking about something rather different than the original concept. That’s what I mean by veering a little off-base.

In any case, “fight, flight, freeze, or fawn” is now the oft-cited list. Proponents of functional freeze present their innovation as an outgrowth of one of these possibilities, lending it a scientific veneer.

Unfortunately, this concept veers much further off-base. It’s a contradiction in terms. A deer caught in the headlights doesn’t “feel” frozen while going on with its usual activities. A person scared stiff by an attacker in a dark alley doesn’t apathetically buy some groceries in that state. Frozen means frozen. Functional freeze bears no resemblance to an acute stress response like fight, flight, or even freeze.

Proponents wave this away by saying functional freeze is not an acute stress response, like a deer in the headlights, but a reaction to chronic stress. They say it’s a compromise “rather than the full paralysis of a classic freeze response.”1

This begs questions about the alleged compromise, including how we compromise with this stress reaction when we don’t, and often can’t, with others. In short, it’s not really a coherent idea.

Isolation of affect

There are more helpful ways to understand this kind of state. We needn’t torture the idea of an acute stress response to make it fit non-acute situations. And we needn’t ignore decades of psychological understanding in order to account for functioning while feeling unmotivated, disconnected, or numb inside.

First, we should note that we’ve known about such states forever. Going through the motions with a vague sense of disquiet, without your heart being in what you’re doing? Shakespeare wrote detailed, nuanced character studies of this over 400 years ago. In the 19th century the same state was called neurasthenia. It was a fad diagnosis back then, even more than functional freeze is now. Then Freud noticed it in the 1890s and called it isolation of affect.

Ever since, psychoanalysts and psychoanalytic therapists have recognized isolation of affect as a defense mechanism. Like other defense mechanisms, e.g., denial or projection, isolation of affect protects the subject from disturbing feelings by keeping them out of awareness.

Picture a person who witnessed a horrible car accident but later describes it in a monotone voice. They were there, they can say what happened, but their mind desperately avoids the feelings connected with it. Or picture someone who goes to the funeral of a close relative or friend, but feels numb and disconnected while there. They’re defending against their feelings.

These are extreme examples, but minor ones occur all the time. There are many things we agree to do, or realize we have to do, but our heart isn’t in it, or we plainly don’t want to do it, or doing it makes us really uncomfortable. So we go through the motions—we do what we have to do—while blocking out our feelings about it. We isolate our affect.

This accounts for functioning that is listless or apathetic, not truly normal. Our emotions are propulsion, the driving force that gets us to do things. Without the emotional charge—even if that emotion is negative, like anger or fear—a thought such as “I need to clean the house” or “I need to do my taxes” is just a dry fact. It’s not motivating. You may still act, but in an unmotivated, low energy way.

Also, imperfect defense mechanisms trigger some degree of anxiety. It’s the psychic price we pay for keeping things from ourselves.

Unfortunately, isolation of affect can look like laziness: slogging through life, doing the bare minimum without enthusiasm. Social media influencers spare sufferers the derogatory “lazy” label by substituting “functional freeze.” It’s a worthy aim, but they resort to a meaningless term and concept. In many instances, isolation of affect can account for what’s going on.

Learned helplessness

However, isolation of affect alone doesn’t address chronicity or pervasiveness. Why would a person feel emotionally numb and defeated day after day, across different aspects of life? For that, we can add a principle first elucidated by experimental psychologists 60 years ago.

Studies of learned helplessness began with a series of famous experiments by the psychologist Martin Seligman starting in the 1960s. Seligman showed that animals he subjected to inescapable electric shock would later fail to save themselves even if escape was possible.

For example, dogs first suffered repeated electric shocks. Some dogs were in a group that could control the shocks, others were in the group that couldn’t. All of the dogs were then shocked in a similar way, but could escape by jumping a low barrier. Seligman found that the dogs that previously lacked control over being shocked had learned they were helpless and didn’t try to escape when they could. (Experiments like these are now considered unethical owing to the suffering of the animals. They weren’t seen as unethical at the time.)

Similar experiments were done on people as well, substituting loud noise for electric shock. Whether done to animals or people, the result was that subjects exposed to inescapable noxious stimuli did not escape later when they could. Resigned to their plight, they showed low energy, low motivation, and apparent depression. The situation had taught them that they were helpless and that escape efforts were pointless.

Dogs in these studies would lie down passively and whine. Humans likewise reported feeling frustrated and stressed, unmotivated, and inefficient. In later human studies, the subjective picture included hopelessness and demoralization, low self‑esteem and self‑blame, cognitive fog with difficulty learning from success, and emotional blunting on the surface (appearing numb or disengaged) with high physiological stress underneath—almost exactly what people now call “functional freeze.”

The conclusion? If you’re living a life where the bad parts seem inescapable, where you’re subjected to electric shock or loud noise—or a noxious work environment, family dynamics you can’t escape, a life situation that you don’t want but don’t see a way out of—then it’s no surprise you’ll feel helpless. Like those dogs, you’ll lay down passively and whine. Or like the human subjects, you’ll function, more or less, but feel emotionally numb, disconnected, or stuck internally. You’ll have an ill-at-ease feeling of emotional paralysis despite appearing productive. That’s how we react when faced with chronic inescapable stress.

Isolation of affect, first described in the 1890s, explains how we defend against our own painful, unwanted emotions in particular situations: when we witness a car crash, attend a funeral, or have to do something we’d rather not. It’s not an acute stress response, not physiological in the “fight or flight” sense. It’s psychodynamic. But pop psychology now rejects psychodynamics in favor of “trauma”—we’re all just having normal reactions to abnormal situations. This view may be simplistic and false, but at least it doesn’t “blame the victim.”

Feeling numb, low-energy, and unmotivated as a way to protect yourself from an immediate unwanted feeling is isolation of affect. Learned helplessness accounts for apathy and low motivation if you’ve been ground down chronically by your environment.

We can even consider whether these are one and the same. Perhaps some of the subjective feelings accompanying learned helplessness (in people, not dogs) are an adaptive use of the isolation of affect defense. Something to ponder.

Psychotherapy (or psychoanalysis) is the traditional treatment for isolation of affect, mainly by providing a safe place to let real feelings out. Journaling, or just checking in with yourself and being honest about how you feel, can also help. Dr. Seligman found that learned helplessness could be relieved by giving animals or people small, achievable successes, basically to re-learn that they could affect their environment and make their own situation better. Both of these take time, but they’re a way out.

For another unrecognized defense mechanism, see my post on boredom. Like apparent laziness, it’s not always what it seems.

An unruly character barges in and disrupts the surrounding peace and quiet. Loud, gruff words and ill-considered behavior mar the scene. Onlookers cringe at the impending destruction. They know this beastly fellow is bound to break something: upend a friendship or family gathering, make a workplace intolerable. It’s worse still if the setting harbors sensitive souls whose feelings are easily hurt.

We may say this person is a “bull in a china shop.” In this metaphor a powerful animal threatens fragile items. Its untempered impulses—hunger, lust, anger—may bring the edifice crashing down at any instant. Even the natural movements of a calm bull may clumsily destroy order and beauty all around.

The message is clear. This bull needs to be controlled, tranquilized, restrained. As a last resort, we must chase it out of the china shop without delay, before it does more damage.

This metaphor fits some interpersonal situations very well. However, in my psychotherapy work I’ve repeatedly encountered a mirror image of this scenario. The phrase rings differently when the bull isn’t the culprit.

“Bull in a china shop” normally implies that the china shop was there first. The bull wandered in uninvited. But suppose we set up the scene another way. Picture a bull grazing in an open field. Yes, it’s a big powerful animal, and maybe it’s a bit clumsy. But it isn’t hurting anyone. It is living in peace.

Then imagine someone sneaks up on this bull and quickly builds a china shop around it. The animal suddenly finds itself constrained, unable to behave naturally without hearing the crash of broken dishes. Its movements are now destructive, as the china is surely at risk. Yet it isn’t quite right to blame the bull.

In human relationships, the person with visibly disruptive behavior isn’t always the primary problem. This arises most obviously in work with children, who may express parental distress and underlying conflicts through their own misbehavior. In family therapy, such a child is termed the “identified patient”: the individual with emotional or behavioral issues that draw the family’s focus, but whose symptoms actually reflect broader family dynamics and conflicts.

Needless to say, this occurs in adults too. In therapy, adult patients often build a case in calm, reasoned tones that their partner, close relative, or coworker is unruly, uncaring, even beastly. They describe ruffians who threaten them emotionally, who traumatize them without cause. It can take weeks or months before the patient’s own role comes to light. Here are a few of the many ways an “innocent” china-shop builder may provoke a bull:

• Passive aggression, i.e., subtly antagonizing others until they lash out

• Imposing unreasonable demands, creating resentment

• Intellectualizing, i.e., disguising feelings by being “the reasonable one”

• Scapegoating

• Setting inconsistent or unpredictable boundaries

• Emotionally neglecting

• Intruding with enmeshed over-involvement

• Externalizing, or projecting one’s own anxieties

• Communicating poorly, leading to misunderstandings and frustration

In these and similar scenarios, a patient may present as a faultless victim—“I was just managing a quiet china shop when a wild bull charged in!”—even if in reality they provoked the conflict. As I’ve argued elsewhere, “therapy” that consists solely of advocacy and empathy overlooks by design the patient’s own contribution to their misery, and forecloses the potential for personal change.

No matter which came first, the china shop or the bull, the combination is bad news. When psychotherapists hear about a big animal bumping into fragile dinnerware, sometimes the lumbering, bellowing bull really is the problem. But other times we’re hearing indirectly about cruelty to animals. We need to be sensitive to that, even if the animal in question isn’t our patient.

Sorting out these dynamics is hard, but that’s our job. In particular, it can take a long time to realize the bull was just being a bull, and that the root problem was the apparently innocent bystander who constructed a china shop that the bull was almost sure to topple.

Adapted from my blog archives — SR

This restates and expands on my YouTube video:

Boredom occurs along a situational-dispositional continuum. Some situations are boring for almost everyone: solitary confinement, waiting in a lengthy queue with nothing to do, having to perform a tedious repetitive task. Psychologist James Danckert at the University of Waterloo studies this type of boredom.1 He says boredom “signals the need to explore,” and that it amounts to “looking around for something to do that will satisfy your need to act and be an effective agent in the world.”

But much of the boredom people suffer isn’t easily accounted for by their situation. In our complex, fascinating world, with its endless interests and diversions, people who enjoy all-you-can-eat freedom at this smorgasbord still suffer boredom. What accounts for that?

A paradigm example is the teenager with school obligations, a world to explore, and endless recreational options who nonetheless slouches on the sofa and complains of being bored and having “nothing to do.” Or anyone who spends hours mindlessly scrolling social media, or staring blankly at a tv, not really paying attention, feeling like they’re wasting their time.

Psychoanalysts over the past century see this kind of boredom as a defense mechanism, a way to block painful or unpleasant emotions. That is, a way to hide unwanted feelings, impulses, and wishes from oneself.

• In 1934 Otto Fenichel wrote that boredom is a state of “dammed‑up instinctual tension” that functions to keep wishes and associated feelings out of awareness.2

• In 1953 Ralph Greenson reaffirmed some of Fenichel’s ideas and wrote that boredom is a repression of desire and fantasy.3

• In 1960 Donald Winnicott wrote about the “True Self” (our genuine nature), versus a “False Self” that we present to the world as a defense to get along, or make ourselves acceptable. Individuals may feel unreal, empty or “dead” behind their facade, and struggle with authenticity and emotional fulfillment. This description is phenomenologically close to boredom.4

• In 1993 Adam Phillips framed boredom as “a defense against waiting,” against acknowledging desire.5

These are different ways of saying essentially the same thing: that boredom attributable more to the bored person than to a boring environment serves as emotional self-protection. For a drive theorist like Fenichel, boredom blocks drives. For Winnicott and those who followed, it’s the subjective experience of hiding something from oneself: desire, one’s true nature, etc.

Let’s look at how this shows up in contemporary life.

1. “Doomscrolling”

Scrolling for hours feels unsatisfying and empty—not fun or truly entertaining. It’s vaguely compulsive, and may even feel mildly tense and uncomfortable. A good description of this feeling is restless apathy.

Boredom here is escape. It’s the withdrawal of emotional investment from the real world, often because reality feels too agitating, threatening or disappointing. Prolonged scrolling of social media, or watching tv without true interest or engagement, maintains a low-level buzz of distraction that hides disengagement from life.

2. Retreating to Fantasy

Real life feels boring, but a vivid, active fantasy life takes its place. This imaginary world offers highly compelling rewards missing from real life: power, admiration, excitement, etc.

This pattern suggests inhibited aggression or desire. Intense drives—to conquer, to be loved, to experience thrills—feel dangerous in reality. Repressing them feels flat and boring, but they can be displaced safely into fantasy. Boredom is the price of emotional safety in the real world, while fantasy/imagination is the compromise that allows safe expression of these powerful feelings and urges.

3. Dreading Free Time

When at work or otherwise occupied, there’s no sense of boredom. But unstructured time, say a free evening or weekend, reliably triggers a boredom that verges on anxiety. This is the bored teenager on the sofa with “nothing to do.” Again, restless apathy may be a good description of the subjective experience.

In this case, work or school provides reassuring external structure. But lacking that structure during free time reveals a stark inner world of thoughts, memories, & feelings. If that world is critical or painful, boredom provides defensive insulation. Feeling bored is a way to avoid being alone with oneself.

4. “I Don’t Know What I Want”

This is chronic indifference regarding choice: what to do, eat, read, or watch. A mind that’s blank, without preference, leaves only boredom.

One possibility is that this is defense against interpersonal conflict. A monotone “I don’t care”—a protective bored numbness—sidesteps imagined conflicts with others. This may have roots in childhood, when strong preference or desire caused actual conflict with a parent or other important person. This early lesson taught that “killing” one’s desires, even before they become conscious, is the cost of preserving relationships—even though this is no longer true in adulthood.

These four examples are common, so you may recognize yourself in them. To some degree, we all do. But it helps to be curious about where feelings of boredom come from. In psychotherapy, the moment we stop calling it “boredom” and start calling it “defense” is when the real work begins. So I’ll end with an example straight out of therapy itself.

5. “Nothing to Talk About”

Despite weekly therapy sessions that are usually lively and feel useful, one day the patient comes in with absolutely nothing on their mind. They sheepishly admit they have “nothing to talk about today.” Their mind is blank. There may be an awkward silence. They desperately review the week in their mind, searching for a topic, but only recall it as boring—nothing really happened.

In my experience, these can be the best sessions of all. Why? Because something big and important is being repressed, i.e., avoided. The blankness and boredom are a defense. If my patient can entertain the possibility that it is a defense, we often get to something troubling or unpleasant that’s been eating away at them. Frankly, that’s a lot more useful than a conscious, rehearsed, composed report of how their week went.

As I think about these examples, I’m struck by the restless, low-level anxiety that accompanies each of them. I’m reminded of Freud’s signal anxiety, which he said results when unconscious drives and wishes threaten to enter consciousness. Boredom is a defense mechanism, but like all such mechanisms, it’s imperfect. Anxiety is the reminder that it takes mental energy, it grinds the psychic gears, to fool oneself.

Then again, restlessness also accompanies situational boredom: imagine pacing the floor in solitary confinement, or fidgeting while waiting in line. Boredom is never a state of calmness or rest. Whether due to the setting, the individual’s psyche, or a combination of both, boredom signals that something isn’t right.

Maybe you can think of other ways boredom is a wall that blocks painful emotion. Looking behind that wall is what depth psychotherapy, like psychoanalysis and psychodynamic therapy, are all about.

Underlying commonalities

I delight in linking apparently disparate fields. Years ago I realized that Starling’s Law of the Heart from physiology parallels the Yerkes-Dodson curve from industrial psychology. Starling’s Law says the strength of the heart’s contraction rises as strain on the heart increases1—but only up to a point. Beyond that, excess strain impairs heart contraction. The compensatory mechanism is overwhelmed. Meanwhile, over in industrial psychology, Yerkes and Dodson found that the performance of factory workers increases as physiological or mental arousal increases—but only up to a point.2 Beyond that, excess arousal impairs productivity. Both “laws” reflect a common feature of homeostatic mechanisms: they compensate only within limits. Physiologists can learn Starling’s Law, and industrial psychologists can learn the Yerkes-Dodson curve—or we can all learn about homeostatic mechanisms and apply those principles widely.

As another example, I’m currently working on parallels between two long-time pursuits of mine, psychodynamic psychotherapy and Aikido; they address human connection at a deep level in surprisingly similar ways. But that’s a topic for another time.

Here, though, the fields aren’t really different. Psychoanalysis and social psychology are both psychology. Yet they seem worlds apart.

My last post argued that emotions expressed in psychoanalysis or psychotherapy that are not “common sense”—that are out of place given the actual situation—are the ones most usefully viewed as transference or countertransference.

But what makes an emotion not qualify as “common sense?” It’s when the individual’s response differs from that expected of the larger population, i.e., when individual idiosyncrasies predominate over situational variables that affect everyone equally. In clinical work, we call those individual idiosyncrasies personality. And when they give rise to feelings about one’s analyst or therapist, we call those feelings transference.

The Fundamental Attribution Error

The relative weight of situational factors versus personality factors in explaining behavior was a hot topic in social psychology decades ago. A series of experiments carried out in the 1960s and 1970s were a backlash to then-prevalent psychoanalytic views that ascribed behavior largely to personality. Instead, academic psychologists repeatedly showed that personality was less influential a factor than expected in many circumstances.

In the infamous Stanford Prison Experiment (1971), college students randomly assigned to play roles of jailers and inmates took on stereotypic attitudes of these roles, regardless of their individual personalities. In the Good Samaritan study (1973),3 Darley and Batson found that whether a bystander would help a stranger in distress was largely determined by how time-pressured the bystander was, not by their personality. In a lab “quiz show,”4 questioners were rated as smarter than contestants by subject-observers, despite knowing the roles were randomly assigned and favored the questioner. Observers attributed performance differences to intelligence, a fixed trait, instead of to the obvious situational advantage of the questioner role.

Evidence from these and similar studies led to a principle that psychologist Lee Ross labeled the Fundamental Attribution Error (FAE). It says that in accounting for the behavior of others, we ascribe too much of the variance to the individual’s personality, and not enough to the situation.

Interestingly, this doesn’t hold when we account for our own behavior. If we see a driver run a stop sign, we’re apt to blame enduring personality traits: the driver is irresponsible, careless, a narcissist, etc. This is the FAE in action. But if we run the same stop sign ourselves, it’s because we were momentarily distracted, in an extraordinary hurry, etc.

I spoke yesterday with an avid bicyclist who objected to “distracted driver” language, because it makes recklessness sound transient and situational, not dispositional as she felt it should be. “These are BAD drivers!”

Psychotherapy and the FAE

Psychoanalysis and psychodynamic psychotherapy are tailored to minimize situational influences and maximize personality influences. The latter, after all, are what the clinician works with. The situation is set up akin to a scientific experiment, where extraneous variables are held relatively constant, in order to see the effect of the variable under study. That’s what makes psychoanalysis and related therapies so unlike normal conversation.

Unfortunately, it’s also what invites the FAE. Analysts and therapists may assume their patients’ emotional responses stem from transference, i.e., personality factors, while overlooking situational factors that might better account for them. That’s why I wrote last week’s post, and why social psychologists did all those experiments 60 years ago.

There’s a lot of talk on social media, especially now, about therapists needing to be aware of patients’ social and political realities: discrimination, prejudice, economic constraints, and the like. In other words, we’re cautioned not to fall prey to the FAE. My last post argues this too: don’t ascribe to transference/personality what can be explained more straightforwardly by the “real relationship,” or by the reality of the patient’s situation more generally.

At the same time, I’ve repeatedly criticized a misguided implication (sometimes stated explicitly) on social media, that therapists should frame therapy as political advocacy, validation, and emotional support in lieu of introspection and personal change. That’s not what psychotherapy—not to mention psychoanalysis—is for. We can steer clear of the FAE and still do our job: help patients deal with their world as it is, by treating aspects of their personality that get in the way.

Are transference and countertransference everything each person feels in psychotherapy? Or do these terms only refer to certain feelings? How do we differentiate transference and countertransference that can move treatment forward—feelings or stances that invite interpretation to reveal underlying meaning—from everyday thoughts and feelings we can accept at face value? How do we avoid over-analyzing?

What is transference?

One of Freud’s enduring insights is that a patient’s feelings about their psychoanalyst, a person about whom they know very little, parallel their feelings toward others they don’t know well. And that these feelings, in turn, reveal assumptions or prejudices (or mental models, or interpersonal templates, or “schemas”) about self and other that the patient first developed in childhood. Freud held that feelings arising early in childhood toward caretakers and siblings are “transferred” to the analyst, so he called this transference.

It’s easy to give oversimplified examples of “transferred feelings” that reflect conscious learning. If you were raised by a sadistic father, you may fear older men you meet later in life, expecting them to be threatening or harmful. That’s simple learning: get burned by a stove, and you’re careful around stoves in the future.

Transference can be much more subtle than that. For one thing, it’s often completely unconscious. Maybe you’re a little tense around older men but have no idea why. Or you avoid older men for one or more conscious reasons (that you believe and voice to others), but in reality those rationalizations cover your unconscious fear. For example, you may sincerely feel that older men are cantankerous, or stuck in their ways, or smell funny, and remain completely unaware that your distaste stems from your fearful exposure to sadism early in life. What makes transference a psychoanalytic or psychodynamic idea, and not simple learning (like avoiding hot stoves), is its hidden, unconscious nature that results in unexplained anxiety or other symptoms.

The relative silence of the psychoanalyst or psychotherapist leaves plenty of room for transference. Is this new person, the therapist, a helper or a threat? Are they strong or weak? Supportive or humiliating? Are they a giver or a taker? Psychoanalysts and psychodynamic psychotherapists exercise a degree of restraint in expressing their own personality, so that the patient’s conscious and unconscious assumptions are highlighted.

Interpreting the transference—bringing unconscious transference into consciousness—is the prime tool of traditional psychoanalysis, and it remains an important one in the less intensive offshoot called psychodynamic therapy. But are all of the patient’s feelings toward the treating clinician usefully viewed as transference?

Transference vs the “real relationship”

What I’ll suggest here is a practical way to think about it.

First, a little history. Back in 1967, the analyst Ralph R. Greenson identified three core components of the analytic relationship1:

• the transference, as described above

• the working alliance, the rational, cooperative part of the relationship focused on therapeutic tasks, and

• the real relationship, “the realistic and genuine relationship between patient and analyst…”

Greenson said the way to tell transference from the real relationship is that transference is inappropriate to the present situation and repeats earlier object relations, especially from childhood. This shows up as reactions that are out of proportion or clearly out of step with the therapist’s real behavior. For example, seeing a calm, careful analyst as tyrannical because of an old parental figure. The real relationship, in contrast, is the realistic, two‑way connection in which both parties experience each other more or less as they really are.

In other words, this is largely a matter of uniqueness and degree, or for want of a better term, common sense. If the analyst/psychotherapist fails to appear for a scheduled session, the patient’s angry, hurt reaction is not usefully seen as transference. It’s not specific to that patient—it’s common sense, in that most people would feel that way. Viewing it as transference doesn’t move the treatment forward, nor reveal anything helpful about the patient.

In contrast, if that same anger and hurt follows an extremely minor lapse by the therapist or none at all, it’s likely useful to view it as transference. Most people would not react that way; it’s not common sense. In short, it’s more useful to consider behavior or feelings as reflecting transference when they are unusual or excessive. When they’re not, there’s nothing to interpret.

Seeing transference is always a balancing act. With too stringent a standard, important clinical clues are missed. Too lenient a standard leads to wild goose chases and the patient’s frustration at being “over-analyzed.”

What is countertransference?

The definition of countertransference has invited debate and has shifted over the years. As a first approximation, the term refers to feelings the analyst or therapist has toward the patient. Since therapists are human too,2 countertransference also largely stems from early childhood. When encountering a new patient (or any other person), does the therapist assume they are needy? Deferential? Competitive? Hostile? The recommendation that therapists receive their own therapy aims in part to bring these assumptions into consciousness, so they can be put aside when the person is serving as a therapist. In his public writings, Freud referred to countertransference as purely detrimental; he urged analysts to rid themselves of such feelings by undergoing further analysis.

This view, that countertransference reflects the clinicians’s own personality and is thus an interference, is now termed the “narrow perspective” or “neurotic countertransference.”

But since the mid-20th century, countertransference has been viewed more positively, as the entire body of feelings the therapist has toward the patient (this is sometimes termed the “totalistic” perspective). Countertransference in this sense can be in part a reaction to the patient’s transference, and thus serve as important data for the analyst or therapist to use in treatment. It can serve as a sensitive interpersonal barometer, a finely tuned instrument in the field of social interaction. For example, a therapist’s irritation with a particular patient, seen as clinical data, may point to subtle unconscious provocations by the patient that irritate and repel others, and keep the patient unwittingly lonely and isolated.

Just as Greenson distinguished transference and the real relationship, we need to distinguish countertransference and other sources of feelings in the therapist. To that end, when I teach psychotherapists in training, I suggest they go through a mental checklist whenever they become conscious of possible countertransference:

1. Is this feeling characteristic, i.e., does the trainee have it much of the time, or toward patients in general, regardless of who they are? If so, it says a lot about the therapist, but nothing about their patient. This is “neurotic countertransference,” the kind minimized by the therapist’s own psychotherapy or psychoanalysis.

2. Many feelings arise in sessions that aren’t characteristic of the therapist, but have nothing to do with the patient in the room. Feelings caused by hunger, the therapist’s personal life, the prior patient, bureaucracy in the medical center, and so forth are not useful data for helping the patient. This includes feelings evoked by irrelevant characteristics of the patient, e.g., the patient physically resembles the therapist’s sibling or spouse—aspects that have nothing to do with the patient’s personality or psychology. But see my note below.

3. The classic “real relationship” category: when the feeling is a common-sense reaction to what is happening, e.g., feeling irritation or fear toward a patient who is screaming obscenities and viciously destroying the office. This is countertransference in the totalistic sense, but not particularly revealing or clinically useful.

4. Useful countertransference: when the feeling is not one the therapist has all the time; when it relates to that one particular patient; and when it isn’t a common-sense reaction to an immediately obvious trigger. What’s left are the most helpful feelings to notice as a psychotherapist or psychoanalyst, those that shed light on subtle yet important dynamics in the patient.

Identifying countertransference is once again a matter of uniqueness and degree. If the therapist’s feeling is noteworthy, unexpected, or unusually strong (or weak!), it’s worth being curious about.

As with transference, discerning countertransference is always a balancing act. Dismissing one’s own feelings as extraneous risks ignoring data that can help the patient. A converse danger is mistaking one’s “neurotic countertransference” for useful clinical data. A further complexity is that reveries or intrusive daydreams, e.g., feelings about hunger, clinic bureaucracy, and so forth as mentioned above, may indeed relate to one’s patient. They may serve as a temporary mental escape, or hold symbolic meaning that relates to what the patient is talking about. A sensitive clinician will consider these possibilities, even if the material appears superficially unrelated.

Transference and countertransference are useful enough in treatment that their presence should always be assessed. Nonetheless, everything the patient feels is not usefully viewed as transference, and everything the psychotherapist or psychoanalyst feels is not useful countertransference. Assuming otherwise leads to unhelpful complexity and blind alleys. It may damage the real relationship, the human-to-human connection that also serves a fundamental healing function in treatment. Over-analyzing may frustrate the patient—a common-sense reaction and therefore part of the real relationship—and drive them away. As Ralph Greenson wrote, the real relationship is the backdrop against which transference—and countertransference—can be recognized and interpreted.

What is a Dialectic?

Broadly speaking, a dialectic is tension between two apparently contradictory viewpoints, where a greater truth emerges from their interplay. One classic example is Socratic dialog, in which philosophers who disagree with each other mutually benefit by finding defects in each other’s arguments.

In the early 19th century, Georg Wilhelm Friedrich Hegel described a universal dialectic. His esoteric philosophy is commonly summarized as “thesis, antithesis, synthesis,” although this specific wording actually comes from the Kantian philosopher Johann Gottlieb Fichte.

The concept is that every thesis, or proposition, contains elements of its own negation. Only by considering both the thesis and its contradiction (antithesis) can one synthesize a greater truth. This process never ends, as the new synthesis itself contains antithetical elements. The term veered in meaning with Karl Marx’s dialectical materialism, and in yet other directions with more contemporary writers. But DBT uses the Hegelian/Fichtean sense, and that is our focus here.

Dialectical Behavior Therapy (DBT)

In the late 1970s, the psychologist Marsha Linehan faced a dilemma when she conducted cognitive behavioral therapy with her caseload of chronically suicidal women diagnosed with borderline personality disorder. The treatment itself pathologized her clients, many of whom dropped out of care. Clients thought: “If I need to change, there must be something wrong with me.” To avoid stigmatizing this population, Linehan turned to Zen Buddhism’s self-acceptance and focused on her clients’ strengths. But this, in turn, downplayed their real need to change.

Linehan and her colleagues realized they would need to integrate change (thesis) and self-acceptance (antithesis) into a larger truth that incorporates both (synthesis). They recognized this as a dialectic, one so central to the work that Linehan named her method dialectical behavior therapy, or DBT.

The apparent contradiction of self-acceptance and the need to change is the fundamental dialectic of DBT, although there are others. For example, therapists are trustworthy and reliable, and they make mistakes. Clients are doing their best and want to do better. DBT teaches clients to say “and” (e.g., the therapist is reliable and makes mistakes), not “but.” The aim is to embrace the truth of both propositions at once, not to choose one over the other.

Dialectics Occur In All Psychotherapy

An uneasy tension between self-acceptance and the need for change exists in all psychotherapy, not just DBT. It starts with the decision to pursue psychotherapy at all. Therapy begins only when a perceived need for change contends with the previous acceptance of emotional discomfort. Most clients feel these competing pulls when entering treatment, a dialectic state that reflects a greater truth about the reality of their situation.

Then, once in treatment, discussion often hinges on the client’s daily struggles with change versus acceptance. File for divorce or work on one’s marriage? Learn to be bolder or accept that one is shy by nature? Alter oneself through exercise or plastic surgery, or become more comfortable with the body one has?

At first glance, such questions appear to be resolvable by tallying pros and cons to reach a logical conclusion. And often that works, in which case no psychotherapy is needed. However, sometimes these questions are surface manifestations of anxiety-ridden inner conflicts that can’t be resolved logically.

Such struggles were first described by Freud, who counseled therapeutic neutrality. That is, psychoanalysts were not to endorse either side of an internal conflict — not choose sides or offer advice about which path to take.

As a practical matter, we rarely know which option is best for the individual in our office. More fundamentally, choosing one side or the other won’t resolve their conflict. The therapist who opines that their client should divorce will prompt the client to argue for staying married. The therapist who advises saving the marriage will prompt arguments for divorce. The conflict persists not for logical reasons, but for unconscious emotional reasons.

Conflicts of this sort aren’t resolved at the level of the concrete opposing options. Instead, psychoanalysts and depth psychotherapists point to insight as the cure. But what is insight? Could it be another way of referring to dialectical synthesis?

Creative, Third-Way Outcomes

Dialectics goes further than simply resigning oneself to irresolvable dilemmas. A clash of thesis and antithesis may result in a new third way, a synthesis that incorporates, yet transcends, both sides of the argument.

This “union of opposites” was first described by pre-Socratic philosophers, and shortly thereafter by Taoists roughly 2500 years ago. About 1000 years ago, the Taoist concept was codified in the famous (“taijitu” style) yin-yang symbol of interdependence: the shapes are opposite yet interact to form a perfect circle.

Note, too, the contrasting dot within each color. The notion that every thesis contains elements of its own negation was represented by the Taoists centuries before Hegel.

The union of opposites concept was adopted by alchemists in China, India, and the West. Medieval European alchemists observed that compounding two dissimilar chemicals can result in a third unlike either parent. For example, sodium, a highly reactive metal, plus chlorine, a poisonous gas, produces table salt.

Carl Jung studied alchemy and weaved the union of opposites into many of his psychological writings. It forms the basis of his “transcendent function” that leads to psychological change:

The shuttling to and fro of arguments and affects represents the transcendent function of opposites. The confrontation of the two positions generates a tension charged with energy and creates a living, third thing… a movement out of the suspension between the opposites, a living birth that leads to a new level of being, a new situation.1

Creative, “third-way” processes in therapy are not limited to Jungian analysis. All psychotherapy that isn’t hamstrung by rigid goals and preconceived outcomes plays host to unexpected, apparently serendipitous resolutions. Such treatments embrace thesis and antithesis (intrapsychic conflict) to synthesize the unanticipated and unpredicted, a transcending of the limited options that were first apparent.

When caught on the horns of a dilemma, the solution is not to grit one’s teeth and choose an option. The other will immediately assert its validity. Instead, step back and accept the validity of both positions: It is valid to seek autonomy and relatedness; to be serious and to play; to feel self-pride and shame; to love and hate another person; to feel hope and despair. And it is certainly valid to accept oneself while also striving to change.

Insight is the term used in psychoanalysis and related schools of depth psychotherapy for achieving synthesis: a position that reconciles and transcends thesis and antithesis, feels true in an emotionally deep way, and works in one’s life. We do not progress in our emotional or spiritual development by discarding inconvenient or contradictory truths. Dialectical tension generates creativity and psychological growth, in psychotherapy and beyond.

Adapted from my blog archives — SR

The Promise and the Peril

We risk making two mistakes with AI: shunning it unthinkingly due to fear, or adopting it unthinkingly due to hype. Artificial intelligence is already changing our lives for the better — and for the worse. Can we steer a careful middle course, where we reap its significant benefits without hurting or destroying ourselves?

The full scope of this question for our culture, and for the larger world, can make one’s head spin. But medical practice, and more narrowly mental health care, is a representative domain where this tension looms large and challenges us. Perhaps lessons from my field can shed light more widely.

AI in Medicine

On January 19, 2026, the New York Times published a “guest essay” (their term for an op-ed) by Robert Wachter MD. It was titled, “Stop Worrying, and Let A.I. Help Save Your Life.” Wachter, a prolific internist who has published six books and 300 scientific papers, chairs the top-rated department of internal medicine at the University of California San Francisco. The essay is a promotional piece ahead of his latest book, A Giant Leap: How AI Is Transforming Healthcare and What That Means for Our Future. It’s due out next month.

In his essay, and presumably in his book, Wachter emphasizes how AI can transcribe and summarize doctor-patient encounters, boil down voluminous medical charts, and compile curated lists of possible patient diagnoses that doctors can then weigh using their own clinical judgment. On his website, the blurb for his new book says AI can also “… draft notes, field patient questions, recommend treatments, interpret images, and guide surgeries.”

None of these uses is without risk. AI that ingests private medical information must be “HIPAA-compliant.” That is, it must abide by the Health Insurance Portability and Accountability Act of 1996, a federal law that governs the privacy of personal medical data. The popular consumer versions of AI chatbots such as ChatGPT, Claude, and Gemini are not HIPAA-compliant, while dedicated medical apps certainly should be. Unfortunately, HIPAA-compliance is not an ironclad shield against hackers and other data breaches. So the first and most obvious risk in using such tools is potential loss of patient privacy.

There is also a real risk of mistakes, often called AI hallucinations. During a casual chat at home, the occasional bizarre response from Claude or Gemini usually evokes laughter, not concern. But what if this occurs when fielding a patient’s medical question? Or when recommending treatments, interpreting images, or guiding surgeries in a doctor’s office?

Medical applications of AI require safeguards against hallucinations and similar glitches. Although error-correcting technology is improving rapidly, currently the only safe way to manage this risk is human oversight by experts who know an error when they see one. Medical AI should not run unsupervised when patient care is on the line.

This points to an important distinction that Wachter fails to make in his essay (perhaps he makes it in his book): the difference between using AI as a supervised tool in medicine, versus using AI to treat patients directly, without oversight.

An example of the latter is a pilot program in Utah, where the startup Doctronic uses unsupervised AI to renew “routine” prescriptions for patients with chronic conditions. (Coincidentally, Doctronic’s co-founder is a UCSF associate professor, although not in Wachter’s department.) Wachter doesn’t say whether we should “stop worrying” about applications like this in internal medicine.

AI in Mental Health

He’s much more clear that we should stop worrying about unsupervised AI in mental health care. He chides readers for “disproportionately focusing on rare bad outcomes” regarding driverless cars and “A.I.-based mental health assistants.”

Here he’s not referring to AI assistants that take notes, summarize charts, or generate lists of possible diagnoses under human oversight. He means unsupervised AI furnishing clinical services, akin to Doctronic: “… millions of patients are now able to receive counseling via bots when a human therapist is impossible to find or impossibly costly.”

These “AI-based mental health assistants” aren’t assisting clinicians. They’re assisting patients, for good or ill, by replacing those clinicians.

“AI-based mental health assistants” come in different types. There are tools to assist mental health professionals in our work: the scribes, note-takers, summarizers, and the like. With privacy safeguards and with careful correction of hallucinations and other errors, carried out by the software itself and by human oversight, these innovations are valuable and constructive.

Another species of “AI-based mental health assistant” is less common, but may also enhance certain types of practice. These are apps used by patients between sessions with their human therapist. The apps may perform automated mood assessments, serve a check-in function, or offer psychoeducation. These apps, too, literally assist a human therapist.

However, the most popular example by far of “AI-based mental health assistants” is the public’s unstructured and unsupervised use of general-purpose AI chatbots for encouragement, validation, and the like. These chatbots are not HIPAA-compliant and do not protect users’ privacy. Anything typed or spoken into the chatbot can be reviewed by computer personnel at the company, and even used as training material for the next generation AI.

These chatbots also have minimal safeguards (at best) to protect those in emotional crisis. Their very design, to maximize user engagement through sycophancy and agreement, contributes to widely reported “AI psychosis” and can encourage suicidal users to take their own lives.

Chatting is Not Psychotherapy

Chatting with a bot is also not psychotherapy. “Therapy” was once shorthand for “psychotherapy,” a series of verbal interactions guided by theories of psychopathology and its treatment. Although it occurs within the framework of a human relationship and may superficially resemble everyday conversation, psychotherapy is a treatment strategy conducted by a trained professional. It has a therapeutic goal: to change something within the patient or client in order to relieve emotional suffering or dysfunction.

“Therapy” no longer means any of this. Now it often stands for encouragement, validation, political advocacy — basically anything that feels good emotionally, situates the problem in the outside world, and doesn’t require the patient to change. Chatbots like ChatGPT, Gemini, and Claude play right into this. They have no therapeutic goals — they have no goals of any kind, except to keep the user engaged —and tend to agree with the user’s externalized framework (e.g., “I’m fine, the problem is my spouse.”) They offer “therapy that isn’t psychotherapy.”

When Wachter writes that “millions of patients are now able to receive counseling via bots,” it’s obvious that his idea of counseling is “therapy” in its current loose sense. True psychotherapy is nowhere to be seen.

Dedicated AI Therapy Apps

What about AI apps specifically designed to offer psychotherapy? Some have been shown empirically to decrease users’ depression and anxiety. As a rule, they are HIPAA-compliant and do have treatment goals and strategies. No need to worry about these at least, right?

Not so fast. First, in addition to potential data breaches, they still risk AI hallucinations and other errors. These apps are marketed to replace human therapists, not assist them. They are designed to operate without human oversight. That’s a warning flag: as in medical practice, AI intended for mental health treatment should not run unsupervised when patient care is on the line.

Yes, the public “disproportionately focuses on rare bad outcomes” by “AI therapy.” That’s not hard to understand though. Such mistakes are novel and newsworthy. Also, we accept “to err is human,” but feel that life-threatening errors by machines are avoidable. They’re someone’s fault. Why can’t we build driverless cars that don’t make mistakes? Why can’t we make AI that doesn’t encourage teenagers to kill themselves?

Let’s put that aside for now, for even with hypothetically flawless data security and error-free operation, the drawbacks of “AI therapy” go much further. First, such therapy only comes in one flavor so far: cognitive behavioral therapy or CBT. Cognitive behavioral therapy is not the “best” type of therapy, not the “gold standard,” not the approach with the widest applicability. It’s simply the easiest kind to program. And unlike a human therapist, AI can’t alter its technique if a different approach is needed, nor refer the patient to someone (or something?) that offers what it cannot.

Carefully constructed AI-delivered CBT is far better than nothing. Unfortunately, it is also far worse than CBT conducted by a competent human psychotherapist. A human clinician picks up on subtle facial expressions, tone of voice, speech hesitations, even things like posture and hygiene, to get a global sense of the patient or client — and to tailor what the therapist says and when to say it. AI can’t do any of that. And again, the lion’s share of AI for emotional support and “therapy” consists of ChatGPT and its competitors. That’s the “millions” of users receiving “counseling via bots.” There’s hardly any carefully constructed AI-based CBT in sight.

Assistance Not Replacement

The US health care system is a confusing, inefficient, expensive mess. If AI technology can lighten the load, sharpen medical decision making, make doctors and nurses happier at work, help patients make appointments and answer their questions, enhance medical student teaching… then wonderful, bring it on. All the same applies to mental health care too, of course.

However, blurring the distinction between AI assistants and AI replacements is intellectually dishonest. Humans remain better at interacting with other humans. We’re more adaptable, pick up more subtle interpersonal cues, are more sensitive to nuance. Humans connect to humans in a unique, hard-to-define way. Especially in the field of mental health, but also in health care as a whole, these subtle distinctions make all the difference. AI hype — our fascination with shiny, cutting-edge tech that coincidentally stands to generate billions of dollars — seduces us to trade away our humanity too cheaply.

Given economic pressures to automate healthcare, and the level of AI hype in our culture, not enough of us are worried about AI’s inroads. We shouldn’t halt technological progress when it truly serves us. But we need to weigh the price we pay in quality of care, in turning intimate human healing into a mechanized commodity, when these exciting tools start replacing, not just assisting, human-centered health care.

Zooming out, health care may be the canary in the coal mine. Consider how many other industries and human services will be faster, cheaper — and colder — as AI replaces human interaction. AI tools will revolutionize society in ways we haven’t yet conceived. As it sweeps through our lives, let’s not allow it to revolutionize our humanity away as well.

As to time, I follow the principle of payment for a fixed hour exclusively. A given hour is assigned to each patient, and that hour is his and he is responsible for it even if he does not make use of it. This practice, which for the music or language instructor is considered normal in our society, when it involves a physician sometimes appears harsh or unworthy of his role…1

Late-cancellation and no-show fees apply far beyond psychoanalysis, and now exist in many dentist offices, hair salons, restaurants, hotels, and spas. Many airline tickets and college courses are nonrefundable, as are season tickets at the symphony or ballpark. The rationale in all these settings is that another patient, client, or customer cannot immediately take the place of a cancellation or no-show. The time and resources of the doctor or business have been wasted.

Freud’s successors modified and refined this policy in different ways. At one extreme are psychoanalysts who still charge for any missed session, planned or unplanned, regardless of reason or advance notice. The analyst announces his or her vacation dates and holidays well in advance, and patients may choose to plan their own accordingly. A more lenient if less clear-cut approach is to waive the fee if the therapist can fill the hour with another patient. More commonly, therapists waive fees for sessions cancelled with advance notice. The amount of required notice is specified beforehand and varies considerably among clinicians. The American Psychiatric Association’s code of ethics cautiously endorses this approach:

It is ethical for the psychiatrist to make a charge for a missed appointment when this falls within the terms of the specific contractual agreement with the patient. Charging for a missed appointment or for one not canceled 24 hours in advance need not, in itself, be considered unethical if a patient is fully advised that the physician will make such a charge. The practice, however, should be resorted to infrequently and always with the utmost consideration for the patient and his or her circumstances.

It’s unclear to me, by the way, what “resorted to infrequently” means. Only for carefully selected patients? Only occasionally for a given patient? What is the approved policy the rest of the time? On the contrary, it seems more fair and more conducive to a clear therapeutic frame to “resort” to any such policy consistently, not infrequently.

Which leads to one terminology note before proceeding. I refer to “the therapeutic frame,” a central if somewhat abstract concept in psychoanalysis and related psychotherapy. The “frame” includes all the factors that provide consistency, predictability, and safety for the patient, and thereby allow emotional vulnerability and the lowering of defenses. Fee policies are one aspect of the therapeutic frame.

In any case, under all three of the policies mentioned so far, the reason for the absence has no bearing on whether the fee is charged, although obviously it can be discussed and explored in the therapy itself. Conversely, some therapists don’t charge even for uncanceled no-shows given a compelling reason. Since many psychiatrists and other therapists have policies that differ from the APA ethical standard and from each other, it is fair to say there is no consensus in the field about these policies. Here’s a survey of various options, with my reflections on each.

Never waiving the fee

There is a certain cold logic to the Freudian standard of never waiving the fee for any reason. That is, charging the usual fee whether the analysand (patient) shows up or not, regardless of advance notice or reason. In addition to precluding lost income, this policy underscores the notion that psychoanalysis is an ongoing relationship, not merely a series of individual sessions. It also provides a very consistent therapeutic frame, in that subjective judgments of the analyst never enter the picture. When analysands fall ill or are forced to remain at work during their analytic hour, they may pay the fee with gratitude that the analyst is holding “their” hour, pay with some regret, or pay while bitterly railing against the autocratic, unfeeling analyst. However they react it’s all transference, and interpreting transference lies at the heart of psychoanalysis.

Well, not always. For analytic theory also recognizes the “real relationship,” distinguished from transference and the working alliance by analyst Ralph Greenson in 1967. The real relationship includes the practical realities of two people engaged in analytic or psychotherapeutic work. Many would argue that never waiving fees, regardless of circumstance or even months of advance notice, is not very realistic for the world we live in. That’s my view, too. For one thing, this policy strikes much of the public as unreasonable, thereby discouraging potential patients from entering psychoanalysis at all.

“Filling the hour”

The next contender, to waive the fee if the therapist can fill the hour with another patient, is apparently popular among psychoanalysts, although in my experience it’s less common among non-analysts. Since the hour is paid one way or another, this policy, too, guarantees that income will not be lost. However, in this case the outcome for the patient hinges on the analyst’s behavior, i.e., whether and to what extent the analyst attempts to fill the hour. Since the reality of these efforts, and therefore the actual likelihood the fee will be waived, are unknown to the patient, this approach also invites a wide range of transferential fantasies: that the analyst strives tirelessly to fill the hour, or couldn’t care less; has no other patients, or has a long, eager waiting list; is meticulously honest, or charges the fee regardless of actually filling the hour; and so forth. These reactions can usefully shed light on the patient’s dynamics, moving the treatment forward.

The drawback is that this banks on eroding the therapeutic frame. A policy of “filling the hour” evokes transference in a manner akin to other unilateral disruptions by the therapist. Psychoanalysts and psychodynamic therapists aim to avoid disruptions, such as arbitrarily changing the length or frequency of sessions for a given patient, even if doing this would prompt clinically useful transferential fantasies. The cost in trust is too great. On the contrary, consistency provides the container within which emotional vulnerability can occur. Likewise, waiving the fee for a canceled session should not depend on how busy, diligent, honest, or popular the analyst really is. These are outside the patient’s control, leaving any outcome feeling arbitrary from the patient’s perspective. Whether a psychoanalyst or psychotherapist charges a fee in a particular instance should depend on patient factors, not analyst factors.

Advance notice policies

The most typical policy in psychotherapy practice is for the therapist to announce at the start of treatment how much advance notice is required to avoid being charged for a cancelled appointment. This can range from the 24 hours suggested in the APA code, to two weeks or longer. In my experience, one or two business days is typical, although some therapists require notice by the previous session, often a week earlier.

This policy enjoys the therapeutic-frame advantages of consistency. The patient knows, based on his or her own behavior, whether a fee will be charged. This is analogous to knowing that psychotherapy starts and stops on time, such that if one is X minutes late, there are Y minutes left for therapy that day. The disadvantages are that cancelled sessions may result in lost income for the therapist, and that no distinction is made between frivolous cancellations (where the fee is still waived if announced well in advance), and dire emergencies (where the fee is usually charged, since such absences are often unanticipated).

Of course, therapists can break their own rules and refuse to waive the fee for a frivolous cancellation, or waive it for a sudden emergency. The advantages of consistency are lost — traded away, in effect, for the “real relationship.”

A related challenge is that advance-notice policies permit edge cases. For example, the patient of a psychotherapist who requires 24 hours notice may repeatedly cancel 25 hours in advance. Policies that work well when invoked occasionally may not when invoked repeatedly. Seemingly frivolous cancellations, or repeated cancellations that fall just within one’s policy parameters, present a clinical dilemma, and often evoke negative countertransference. The therapist may feel the patient follows “the letter of the law” but violates its spirit.

To circumvent this, one might opt for a policy conservative enough that even edge cases aren’t irritating. Suppose a 24-hour policy feels adequate for rare absences, but 48 hours feels adequate for repeated absences. The therapist may establish a general “48 hours” policy to allow for this possibility. This aims to avoid negative countertransference, but does so at the cost of unnecessarily restricting (and/or charging) most patients to avoid being irritated by a few. It also does nothing about frivolous absences. And the truth is, it can’t really prevent negative countertransference.

A better option is to impose a new, more restrictive policy on a case-by-case basis. For example, a therapist may establish a new 48-hour policy for the one patient who repeatedly cancels 25 hours in advance. Doing so is analogous to inviting all patients to call after-hours in case of emergency, but restricting this privilege for a particular patient who calls a half-dozen times every night.

This approach accepts the possibility of negative countertransference and treats it as useful data. A patient who tests limits, even without literally violating them, may benefit by exploring the dynamics, even if invited to do so by their mildly irritated therapist. However, as usual there’s a counterweight: changing policies for a patient mid-stream may feel unfair. The patient has “sunk costs”: financial and emotional investment in a psychotherapy that is no longer “what they signed up for.”

Grappling with such issues illustrates the subtlety of balancing humanistic leniency and hard-edged limit setting. Swing too far to the former, and one’s own misgivings will lead to “unfair” mid-course corrections — or simmering resentment, which may be worse. Swing too far to the latter, and one’s treatment will lack human connection and warmth. Since it’s not possible to anticipate how a therapy will unfold, the treatment frame is always an approximation, in dynamic tension between these two poles.

Alternative policies

Here are some policy alternatives aside from those mentioned so far.

One is to decide, on an event by event basis, whether to waive the fee depending on the reason for the absence. This strategy pits the therapist’s values against the patient’s, establishes a dynamic of judging the patient, and, in effect, metes out punishment when the patient’s rationale is “not good enough.” I can find little to recommend it.

Another option is never to charge for missed sessions. Whether the patient calls five minutes beforehand or not at all, if the session doesn’t happen, there’s no fee. This is financially untenable for the great majority of us who see patients or clients by appointment and make our living this way. It’s also the wrong message to give to patients: that psychotherapy is a casual encounter rather than a professional service, and that it doesn’t matter to us, or to the integrity of the treatment, whether they show up or not. On the contrary, we aim for a therapeutic frame, including fee policies, that promote consistency and participation in the process.

Is it possible to have no policy at all? With each canceled or missed session, the two could discuss whether the fee will be charged. This risks defaulting to the subjective judgment policy mentioned above, where the therapist ultimately decides whether the patient’s reason for cancelling was “good enough.” It could also mire the treatment in endless discussion (or arguments) about cancellations and fees. It may not provide a sufficient therapeutic frame; that is, it may be too anxiety-provoking for both parties.

Little wonder I’ve never heard this idea seriously proposed. But handled cautiously and sensitively, a non-policy would underscore the collaborative, co-constructed nature of therapy. And “endless discussion” is what psychodynamic therapy is largely about anyway, assuming the presence of a clinician dedicated to avoiding argumentative enactments.

What I do

My own policy is to waive fees for sessions canceled at least a day in advance. I rarely make exceptions. As a policy it is not particularly onerous, and patients seem to understand that I cannot realistically fill a suddenly vacated hour, even one canceled for good cause. When patients cancel sessions only a few days in advance, I sometimes fill the hour and sometimes cannot, but I consider that my problem, not theirs. I feel this policy works fairly well for everyone involved. However, it isn’t perfect, as illustrated by this true story:

A patient called on the morning of her appointment to report a bad cold. She was willing to come in that day; however, she wondered if I might prefer to see her later that week when she would be less contagious.

It was an interesting twist on the typical same-day cancellation. In truth, I did prefer to delay her visit. I had a suitable free hour later in the week, and didn’t want to catch her cold. By asking/allowing me to decide, and since it worked to our mutual benefit, I obviously would not charge her for missing that day. We met at the rescheduled time, and all was well.

Yet I wonder. By solving this problem for both of us, i.e., agreeing to reschedule her at no charge, did I make a decision that really was hers? Assuming she was in insight-oriented psychodynamic treatment, would it have been more therapeutic for her to decide between (1) attending her hour while ill, and possibly sickening me, or (2) paying for a missed hour? I leave this as an exercise for the reader.

Adapted from my blog archives — SR

The Emotional Immune System

Your personality and coping skills are your emotional immune system: a complex set of defenses that protect you from psychological threats, much as your body’s immune system protects you from threats like infection. However, both systems can malfunction with allergic reactions and autoimmune diseases, when the defenses designed to protect you ends up harming you instead.

Yes, I’m drawing an extended analogy. Hang tight.

Your Body’s Immune System

Your physical immune system detects foreign invaders, such as microorganisms (e.g., bacteria and viruses), parasites, and toxic substances. It encircles and isolates them, then inactivates them, rendering them harmless. We couldn’t survive in the world without this immune system. We’d soon die of overwhelming infection.

Consider too, that some of us have strong immune systems that easily fight off these threats, while others have weak (or weakened) immune systems that can’t protect as fully.

And many of us suffer allergies. Specific threats, such as bee stings or peanuts, trigger an exaggerated immune response that is dysfunctional and even potentially dangerous. Allergies are a cure worse than the disease: a protective function activated when it isn’t needed, bringing adverse consequences that outweigh any possible benefit.

Last but not least are the autoimmune diseases. These include psoriasis, rheumatoid arthritis, lupus, and many others. These conditions result when the immune system, evolved and refined over millennia to detect foreignness, goes off-track and begins to attack healthy tissue as foreign. Such diseases are a challenging medical problem, as any treatment must balance our need for defense against infection against the manifest danger of self-harm. Much contemporary medical research seeks new and better ways to regulate the immune system, in order to mitigate autoimmune diseases.

Your Emotional Immune System

The emotional immune system is similar in all these respects but one. As best we understand, each person’s unique personality begins with an innate, inborn temperament. This is the set of emotional qualities we are born with: our general level of anxiety, our inclination to engage with the world versus withdraw, and so on. These general tendencies appear to be genetic and present at birth. Our innate temperament then interacts in myriad subtle ways with early childhood interpersonal experiences. This process is likely governed by complex nonlinear dynamics that feature interacting feedback loops and perhaps mathematical chaos. This complexity theoretically accounts for the uniqueness of personality and our inability to predict precisely how it unfolds over time.

In part, personality describes the unique way each of us navigates our world of emotions and relationships: how we deal with stimulation, threat, and danger, as well as reward, attention, and love.

Central to personality are the psychological maneuvers, often performed outside of our conscious awareness, that cordon off and neutralize emotional threats, much as our immune system does with microbes.

Although the details differ for each of us, common types of maneuvers are codified into named “defense mechanisms” or “coping skills.” We laugh off uncomfortable situations. We deny unpleasant feelings. We displace our anger away from threatening targets who could retaliate and onto less threatening targets who cannot. We point our attention in directions that are more comfortable for us. We wrap our emotional reactions in layers of intellect, insulating them from awareness and softening their impact.

We couldn’t live in this world, particularly around other people, without these essential features of our personality. That is, without our emotional immune system maintaining safety for our feelings.

Note that neither the physical immune system nor its emotional counterpart is perfect. Viruses still cause colds and flu. Events in our lives still disturb us. We still have symptoms. Nonetheless, both systems keep us out of trouble. They make life manageable.

To extend the analogy, some people have strong emotional immune systems that handle external threats relatively easily. They take challenges in stride, and achieve high scores on psychological measures of global functioning and resilience. We often say that such people have “heathy” or “mature” personalities. For others, this immune system is weaker and emotional threats are more damaging. These people feel overwhelmed by relatively minor stressors, cannot regain emotional equilibrium as quickly, and cannot adapt as easily to changing circumstances in their interpersonal world. We often say that such people have “immature” or “inadequate” personalities, or in extreme cases that they suffer a personality disorder.

Most of us also have “emotional allergies”: specific triggers that set us off, out of proportion to the effect of similar threats, and out of proportion to the reactions others have to those same triggers. Sometimes we call this “pushing our emotional buttons.” Those buttons are a little different for each of us, and are a product of unique vulnerabilities in our emotional immune system. As with bodily allergies, repeated exposure to “emotional allergens” can sensitize the host and trigger an allergic reaction. When someone or something pushes one of our emotional buttons, a protective function is activated when it isn’t needed. This brings adverse consequences that outweigh any possible benefit; it’s a cure worse than the disease.

And last but not least, many people suffer what I’ll call emotional autoimmune diseases. This is when our highly tuned system of personality and coping skills, evolved over millennia to maintain equilibrium in the face of emotional threat, goes off-track and begins to attack parts of ourselves.

Emotional autoimmunity is a challenging psychiatric problem. Any treatment must balance our ongoing need for psychological defense against the risk of self-harm. But here is the single most important difference with autoimmune disease in the body. Unfortunately, very little research seeks new and better ways to regulate the emotional immune system in order to mitigate autoimmune disorders. Current research instead focuses almost wholly on symptom reduction, leaving us largely reliant on older treatment approaches that are time-tested (and empirically validated) but resource intensive for treating the emotional immune system itself.

For the rest of this piece, I’ll compare a real autoimmune disease with emotional autoimmune disease, to illustrate parallels regarding treatment.

Bodily Autoimmune Disease

Psoriasis

What is psoriasis? Inflamed itchy red patches of skin, often with a dry crust on top that flakes off. The severity varies widely, from a single small patch that appears occasionally, to large areas of the body affected all the time. Even though it’s widely seen as a skin disease, psoriasis is fundamentally an autoimmune disease. We diagnose psoriasis when the immune system attacks normal tissue, in this case skin. But it can also attack normal joint tissue, resulting in psoriatic arthritis, as well as the cardiovascular system. And because autoimmune diseases often coexist — after all, the same immune system underlies these different manifestations — psoriasis is a risk factor for other autoimmune diseases, such as inflammatory bowel disease and chronic kidney disease.

In the 1960s, television ads pitched Tegrin medicated shampoo to treat the “heartbreak of psoriasis.” Despite the ad hype, severe cases of psoriasis are truly awful: whole-body itching and disfigurement, bits of skin flaking off everywhere, feeling too self-conscious to wear short sleeves or visit a public swimming pool or beach — not to mention physical intimacy with others. And on top of that, arthritis accompanies skin disease in about 30% of patients. Bad cases can really can be heartbreaking.

Let’s draw a parallel between psoriasis and emotional struggles.

A Comparison with Emotional Distress

Mild and Limited

If you notice a little patch of red, itchy skin, it could be something that’s truly just local, like a mosquito bite. Or it could be a mild case of the whole-body autoimmune disease called psoriasis, in this instance only affecting one small part of the body. Whichever it is, you can treat the immediate problem yourself. Buy a tube of hydrocortisone cream, no prescription needed, or just wait for it to pass.

Likewise, nearly all of us suffer small, contained emotional setbacks from time to time, similar to a small area of itchy skin. If the discomfort is isolated to a single situation or relationship, if it’s self-limited and not too severe, it could be like a mosquito bite. That is, unrelated to anything else, not part of a general underlying problem.

Or it may be like a mild case of psoriasis. That is, a whole-body problem that only shows up locally. We might call this an emotional autoimmune reaction, where your personality, designed to ease stresses from the outside world and keep you safe, is going off-track and hurting you. You feel guilty, blame yourself, inhibit yourself — you’re basically attacking yourself the same way your emotional system would attack an outside threat. In this case, though, it’s just a limited autoimmune attack.

Whether it’s truly an isolated symptom, or a mild manifestation of something deeper, we have many ways to deal with small emotional setbacks. They don’t require a mental health professional. You can call a friend, do something nice for yourself, read a self-help book, talk to an AI chatbot, etc. When the emotional upset is small, we don’t usually worry about whether it’s a mild sign of a more widespread emotional autoimmune disease.

Moderate

What if the problem is a bit bigger? What if you have a few patches of psoriasis on your arms and legs, and the itching really bothers you? Then it may be time to get professional help. A dermatologist may prescribe skin medicine that is stronger than the kind you can buy over the counter. He or she may advise you to get a UV sunlamp, since light treatment helps psoriasis.

Likewise, what if you have an emotional problem that arises in a few different areas of life? Examples include trouble asserting yourself, or being too self-critical. A mental health professional may give you homework assignments, so that you can practice thinking differently, and may give you tips and tools to deal with troubling feelings.

In both cases, these treatments still address surface symptoms. We’re still talking about skin treatments for psoriasis, and treating emotional issues symptomatically, without worrying about where those symptoms came from. Yet at this stage it’s clear that the skin disease is not strictly local, like an insect bite. It reflects an underlying autoimmune disorder. And it’s increasingly clear that the emotional problems do as well.

Severe or Pervasive

What if the problem is even worse? Now the psoriasis is all over your body, and the itching is unbearable. What if you have psoriatic arthritis too? Then it’s time for systemic treatment that goes to the root of the problem. With psoriasis, that means medicine to calm the immune system, so it no longer attacks the skin and other normal tissues.

Note that treatment isn’t focused on the skin anymore, even though that’s the most obvious and troubling place the disease shows up. Systemic treatment indirectly treats the skin, the joints, and any other tissue the autoimmune reaction is damaging. Treating the root of the problem, not its surface manifestations, ends up treating all the damaged tissues. They don’t need to be identified and treated individually.

Turning to emotions, what if you sabotage every relationship, continually undermine yourself, or can’t hold a job because you repel people? What if you’re ruining your life? Dealing with these problems on the surface, problem by problem, is a losing battle — like treating severe psoriasis by focusing only on the skin. The real problem is a full-blown emotional autoimmune disease.

The sufferer’s personality, with its coping skills and defense mechanisms, is turned against the self. The damage is widespread, partly manifest and partly obscure. As in the case of severe psoriasis, the symptom severity and the wide range of dysfunction interferes globally with life satisfaction. And as with severe psoriasis, this calls for systemic treatment.

Depth Psychotherapy

Systemic treatment for emotional dysfunction is called depth psychotherapy. This umbrella term covers several schools of psychotherapy, including psychoanalysis, psychodynamic psychotherapy, gestalt therapy, existential-humanistic psychotherapy, and others. Depth psychotherapy does not focus primarily on surface complaints. It explores the patient’s whole personality. It treats the emotional immune system.

Are idiosyncratic “allergies” at play? Is there an “autoimmune” process damaging healthy functioning far and wide? Depth psychotherapy aims to settle the psychological immune system, so it can deal with outside stress while not attacking the self. In this way it indirectly treats the mood, the relationship struggles, the work dissatisfaction, and other domains that autoimmunity damages. Treating the root of the problem, not its surface manifestations, indirectly treats all the damaged tissues… I mean emotional difficulties. They don’t need to be identified and managed individually.

Depth psychotherapy is the process of taking a deep look into the structure of our own personality: the ways we deal with stress, the ways we see ourselves and others, the ways we manage our feelings. How do these ways end up unwittingly hurting us? Becoming aware of emotional autoimmunity, how we hurt ourselves without meaning to, without even realizing it, is how we begin to ease this heartbreak.

Like most bloggers, I had many ideas at first. I posted every couple of days, then once a week. This eventually lengthened to every couple of weeks, then monthly, and so forth. “Reidbord’s Reflections” never really took off as public education, nor as a referral source for my practice. In recent years I posted rarely, and in fact haven’t posted anything at all since November 2024. I guess I let it die, although it’s still there if you’re interested.

Instead, I launched a YouTube channel in April 2024. I called it “More Than Meds,” which had been my website’s tagline. Besides sounding a hell of lot better than “Reidbord’s Reflections,” the name accurately reflects my interest in all things psychiatric other than the actual prescribing of medications: psychotherapy, ethical and philosophical issues in my field, public perceptions of psychiatry and psychotherapy, artificial intelligence, and so on.

I have nothing against psychiatric medications, by the way. It’s just that the public sees psychiatry largely as prescription-writing, and often for good reason. I hope to counter both the actual reductionist commodification of the field, and the bad press that results.

I’d still love to demystify psychiatry and psychotherapy for the public, and if I’m being honest, attract patients to my practice. But social media, including YouTube, is already a firehose of misleading, oversimplified psych “demystification.” I now see my aim more as debunking, setting the record straight, in the face of rampant memeified paeans to pseudo-therapy — what I call “therapy that isn’t psychotherapy.”

Seasoned YouTubers advise newcomers to post weekly or more to build an audience. So for the first year, I released a video every week. As an inveterate DIYer, I did my own scripting, filming, editing, etc. I still do. Everyone’s first videos are amateurish, and mine were no exception. There’s a learning curve. In the second year, I’ve gotten somewhat better. I also relaxed the pace to a more manageable video every two weeks (sometimes three if I’m on vacation).

If you’re reading this, please check out “More Than Meds.” I invest a lot of time and effort into it, and it will remain my main avenue of outreach and professional expression.

So why am I here on Substack? Two reasons, maybe three. First, I plan to revise and repost a selection of my old blog entries. Some, in my humble opinion, deserve a wider audience. The other reason is that some of the ideas I present in 15-minute videos may benefit by more careful treatment in print. For example, right now I’m working on a video where I draw an analogy between the body’s immune system and our “emotional immune system,” our personality. Obviously, I hope it comes across on video, goes viral, and brings me fame and fortune… TED talks, book deals, the whole nine yards. Or more realistically, an uptick in views. Yet I also imagine the analogy may make more sense if I spell it out here. And a possible third reason is to share half-baked or brief ideas that don’t warrant a video.

Oh, and one last thing. While I always encourage comments on my YouTube channel, I believe this forum is more conducive to thoughtful dialog, the kind of give and take that sharpens everyone’s thinking. I guess we’ll see.

Thanks for reading.

Subscribe now