A man sets himself the task of portraying the world. Through the years he peoples a space with images of provinces, kingdoms, mountains, bays, ships, islands, fishes, rooms, instruments, stars, horses, and people. Shortly before his death, he discovers that that patient labyrinth of lines traces the image of his face.
⏤ Jorge Luis Borges, Dreamtigers, Epilogue
I
Plastic surgery outcomes research literature is much like you’d expect:
Certainly there are factors that influence quality of life that are common to each of these specific interventions. For instance, acceptance by friends and family is an important component of the patient’s quality of life. Similarly, the manner in which the individual’s appearance affects his or her social or professional life is also a common concern. There are also certain common emotional or mental qualities that transcend the satisfaction with any of these procedures. The individual’s confidence and happiness with her appearance, and whether or not she desires some change are qualities that are important components of satisfaction...
The paper[1] goes on to open precisely zero of those cans of worms, and instead proposes a series of short post-operative surveys consisting of six questions regarding the patient’s own satisfaction with their results, as well as their estimations of others’ perceptions of their results. (Again, we’re definitely not opening that can of worms!)
To be clear, I’m not at all against these sorts of methods on principle. As subjective and speculative as 1-5 ratings on questions like “How much do you feel your friends and loved ones like your nose?” may be, some data is much better than no data. If the Bayesian mammogram problem taught us anything, doctors are as susceptible to being horribly misguided by their intuitions as anyone else. As flimsy as a handful of 1-5 scales may seem in comparison to the reams of data collecting in other medical subspecialties, they’ll at least keep us in or near the ballpark and potentially head off common biases that commonly occur with pure guesstimation.
Too, their simplicity is nothing to scoff at the efficacy of simple checklists in medicine is a well-known phenomenon, and seeing that outpatient procedures are harder to collect data on in the first place[2], ease of response is more of a factor.
The best we could do would likely be to subject patients to some gauntlet of psychological evaluations both before and after surgery, not that the typical demographics of elective cosmetic procedures would be keen on enduring added hurdles. And that may just give us more noise than signal: I can’t get around the paywall, but the abstract of a paper titled “Plastic Surgery and Psychotherapy in the Treatment of 100 Psychologically Disturbed Patients” finds that “82% had a positive psychological outcome” and that “there were no lawsuits, suicides, or psychotic decompensations.” This is exactly in line with other studies of more typical patient outcomes, despite the fact that “Patients with psychological disturbances of a magnitude generally considered an ‘absolute contraindication’ for surgery” and were frequently rejected because of the results of their psychological evaluations.
But might we do better with a different angle of attack? With recent advancements in facial recognition and image processing technologies, perhaps a more quantitive assessment could be had. Jürgen Schmidhuber posits[3] that beauty, such as that of a face, is a function of regularity, or “compressibility.” The brain is stingy with its computing power, and thus has evolved to not only the ability to compress sensory data, but the intrinsic reward function to incentivize better ways of compressing data. How might the brain evaluate facial beauty in this model?
For example, to efficiently encode previously viewed human faces, a compressor such as a neural network may find it useful to generate the internal representation of a prototype face. To encode a new face, it must only encode the derivations from the prototype. Thus a new face that does not deviate much from the prototype will be subjectively more beautiful than others. Similarly for faces that exhibit geometric regularities such as symmetries or simple proportions—in principle, the compressor may exploit any regularity for reducing the number of bits required to store the data...This immediately explains why many human observers prefer faces similar to their own. What they see every day in the mirror will influence their subjective prototype face, for simple reasons of coding efficiency.
A neural network could conceivably be set up to either evaluate absolute compressibility, based on perfect symmetry, and before-and-after photographs of patients could be run through it to determine exactly how many bits the face lost relative to perfect compressibility. (In future version of this paper, I hope to have one or both of these built!) Some research has already been conducted on data sets of before-and-after pictures of patients.
Others[4] propose a U-shaped reward function for the brain’s compression algorithm—too regular, and the brain will lose interest; too random, and the brain won’t be able to arrive at any sort of compression at all. Somewhere in the middle lies the most beautiful, for they would have the highest ratio of apparent complexity to compressibility.
A temporal aspect, change in compressibility over time, might also be worth incorporating:
...the human sensation of pleasure that we are trying to explain may well be influenced by a perception of a *change* in compressibility, as opposed to just compressibility. Along these lines, it is not clear whether the subjective experience of music growing on us over time represents 1) a pre-existing cognitive algorithm whose compression potential is only gradually accessed, or 2) an entirely new compression algorithm developed through the challenge and experience of understanding that particular piece of music.
Schmidhuber’s theory is certainly simple and beautiful (for now-obvious reasons), but it’s a little “just so,” and doesn’t reference any neurological sources. Let’s go deeper.
II
The bottom line of James Scott’s Seeing Like a State is that (among other things) top-down systems sometimes work, and sometimes don’t. Allow me to make a superficial comparison: if you had to wager a guess, would you say the brain operates via a jumble of evolved processes, via metis? Or would you say the brain is a top-down autocrat imposing perfect rationality on the various sensors and sub-routines below it?
According to recent neuroscience research, the brain is unexpectedly, more like Le Corbusier and less like Ye Olde organically-evolved village! Predictive processing is something of a grand, unifying theory of the brain, states that two streams of data coexist in the brain—raw “bottom-up” data received from our senses, and a “top-down” stream of predictions based on our high-level knowledge of the world:
To deal rapidly and fluently with an uncertain and noisy world, brains like ours have become masters of prediction – surfing the waves and noisy and ambiguous sensory stimulation by, in effect, trying to stay just ahead of them. A skilled surfer stays ‘in the pocket’: close to, yet just ahead of the place where the wave is breaking. This provides power and, when the wave breaks, it does not catch her. The brain’s task is not dissimilar.[5]
Famed neuroscientist Karl Friston elaborates on the primacy of prediction over perception:
According to active inference, the agent moves body and sensors in ways that amount to actively seeking out the sensory consequences that their brains expect. Perception, cognition, and action – if this unifying perspective proves correct – work together to minimize sensory prediction errors by selectively sampling and actively sculpting the stimulus array. This erases any fundamental computational line between perception and the control of action...Perception here matches neural hypotheses to sensory inputs…while action brings unfolding proprioceptive inputs into line with neural predictions. The difference, as Anscombe famously remarked, is akin to that between consulting a shopping list (thus letting the list determine the contents of the shopping basket) and listing some actually purchased items (thus letting the contents of the shopping basket determine the list). But despite the difference in direction of fit, the underlying form of the neural computations is now revealed as the same.[6]
Remember how I mentioned this is a grand unifying theory of the brain? It really can explain everything: dreaming, priming, the placebo effect, psychosomatic pain, autism, schizophrenia, phantom limbs, motor control. All of these are symptoms of a mismatch between what the top-down predictions of the brain are expecting to see, and what the bottom-up sensory data are actually seeing.
Even if the paper I referenced earlier claiming plastic surgeons operating on mentally ill patients have nothing to worry about, I doubt that any of them would touch one one subset of those patients: those afflicted with Body Dysmorphic Disorder (BDD). There are lots of studies that propose complex psychological screening frameworks for filtering out these patients. These tools are about as effective as you’d expect: 84% of plastic surgeons have unknowingly operated on a patient with BDD and, at least in patients with mild-to-moderate BDD, outcomes for this population are in line with the non-BDD stats (81% positive).[7][8]
But what if BDD is actually predictive processing error? Would that tell us anything about non-BDD patients who are unsatisfied with their outcomes? Would that give us more effective tools than post-hoc psychological narratives in shaping patient outcomes?
III
Cultural theorists often decry the “medicalization” of the body and the instantiation of a “clinical gaze” as a means of cultural control and oppression. That we give this gaze both the permission to deconstruct our bodies into isolated parts, and the sole authority to divine their singular hidden reality, reveals our implicit bias towards imposing arbitrary divisions that may or may not be accurate. Theorists correctly assert that rational systems like modern medicine work because we make them work, not because they are transparent conduits of eternal and absolute truth. By default, then, they engender our cultural ideals, nowhere more so than in cosmetic surgery.
Thus the rituals of the plastic surgeon’s office are suspect, because their medical functions might be auxiliary to their reinforcement of the system they are embedded within. Pre-operation consults are singled out, not as vital preparatory step, but as a type of conditioning:
The female patient is promised beauty and re-form in exchange for confession, which is predicated on an admission of a diseased appearance that points to a diseased (powerless) character. A failure to confess, in the clinical setting, is equated with a refusal of health; a preference for disease.[9]
The rationale for the procedure is also examined. That elective surgeries are performed within the greater medical paradigm of “do no harm” forces an attempt at reconciliation:
Surgeons, who, on the one hand, are keenly aware of the fact that the service they provide is often an entirely elective endeavor, but on the other also realize the potentially serious physical consequences of their medical service. This tension is managed discursively when both physicians and patients construct “curative” justifications for the voluntary submission to surgical treatment...Apparently, the use of “curative” justifications in a diagnosis does not only function discursively to manage an anxious patient, it also legitimates and authorizes the “elective” surgery for insurance coverage.[10]
The dual nature of these practices shaping both outcomes and perceptions of outcomes is seized upon as a contradiction inherent in the very rationality modern medicine derives its authority from. Medicine’s privileged position of objectivity allows them only the former; to shape the latter is a betrayal of what medicine “is.” They’re pulling the wool over our blepharoplastied eyes.
But in the predictive processing model, these rituals, and even “medicalization” itself, might be the most powerful tools we have in affecting outcomes of cosmetic procedures. Instead of giving into the urge to develop more and more sophisticated psychological screenings and evaluations to get at “”what’s really there”“ in a patient, to learn “”the truth”” about an outcome only after-the-fact, what if we took predictive processing seriously and acknowledged that the gradient of clinical outcomes may flow the other direction?
Every book on negotiation tactics I’ve ever underscores the importance of getting people to buy into the negotiation process so that they’re more likely to trust the outcomes. What if each of these “rituals” of the plastic surgeon’s office, each touchpoint in a patient’s journey, from advertising, to before/after photos, to consults, to Photoshop-simulated outcomes, to payment, to the surgeon’s appearance, were studied and reformed with the knowledge that they play just as vital a role in patient outcomes as the physical results of surgery? I’ve found a smattering of research that gives a nod to this line of thought[11] (but even these still maintain that psychological screenings are most effective).
While this might be seen as a exercise in book-cooking, and real medicine doesn’t pull cheap swindles like that and may only use objective means to arrive at “”the truth,”″ consider that the empiricism modern medicine is based on is something of a swindle. Foucault’s thesis in The Birth of the Clinic[12] is that late 18th-century field of medicine was the site of an “epistemological rupture.” Rather than the incremental outcome of a plodding few centuries of accumulating data, the inauguration of modern medicine marked a sharp discontinuity in the field, rendering that which had come before utterly alien and incompatible with what followed:
The clinic—constantly praised for its empiricism, the modesty of its attention, and the care with which it silently lets things surface to the observing gaze without disturbing them with discourse—owes its real importance to the fact that it is a reorganization in depth, not only of medical discourse, but of the very possibility of a discourse about disease.
This is not an isolated trend—legibility was hot stuff for the following few centuries. State planners couldn’t enlist architects with fetishes for straight lines fast enough to help them reboot intuitions onto rationality, onto the “absolute values of the visible.” Some domains stuck the landing better than others. But other than the replicability crisis, which is not a problem unique to the medical field, medicine has benefited tremendously from this shift away from metis and towards “objectivity.”
Even if it’s the case that doctors aren’t able to discern “reality” directly, and operate within an organization of knowledge that varies in its arbitrariness, and said arbitrarinesses can reinforce harmful cultural biases by default, and these biases often go uncontested because everyone assumes medicine to be the irreproachable paragon of rationality, no one would dispute medicine’s efficacy in spite of it lacking “ultimate” foundations and containing inherent contradictions.
Stage 5 recognizes the value of sorting out contradictions within a system, and retains stage 4’s ability to do so. However, it doesn’t expect any system to work perfectly, so it tolerates internal contradictions if they appear relatively unproblematic. Stage 5 entertains multiple systems, and is comfortable with contradictions between them, because systems are not absolute truths, only ways-of-seeing that are useful in different circumstances. Stage 5 is uniquely comfortable with value conflicts, since (unlike both 3 and 4) it does not take any value as ultimate.
Post-modernist critiques like Foucault’s that warn us of inadequacies of building systems on rationality alone are entirely correct. But nihilism, faith in an ultimate lack of meaning, fails for the same reason faith in an ultimate meaning does: reality can stay complex longer than you can stay rationally enlightened. The fact that modern medicine is not based on some ultimate truth should not be the impetus to tear the whole thing down—quite the opposite!:
Rationality works because we make it work—not because it is eternally, absolutely, ultimately Correct. The meta-rationalist agenda asks: what kinds of work, specifically, do we do to make rationality work? How can we do that work better?[13]
"Masks are surrounded by rituals that reinforce their power. A Tibetan Mask was taken out of its shrine once a year and set up overnight in a locked chapel. Two novice monks sat all night chanting prayers to prevent the spirit of the Mask from breaking loose. For miles around the villagers barred their doors at sunset and no one ventured out. Next day the Mask was lowered over the head of the dancer who was to incarnate the spirit at the centre of a great ceremony.
"What must it feel like to be that dancer, when the terrifying face becomes his own?"
⏤ Keith Johnstone, Impro
The Medium Is The Bandage
I
Plastic surgery outcomes research literature is much like you’d expect:
The paper[1] goes on to open precisely zero of those cans of worms, and instead proposes a series of short post-operative surveys consisting of six questions regarding the patient’s own satisfaction with their results, as well as their estimations of others’ perceptions of their results. (Again, we’re definitely not opening that can of worms!)
To be clear, I’m not at all against these sorts of methods on principle. As subjective and speculative as 1-5 ratings on questions like “How much do you feel your friends and loved ones like your nose?” may be, some data is much better than no data. If the Bayesian mammogram problem taught us anything, doctors are as susceptible to being horribly misguided by their intuitions as anyone else. As flimsy as a handful of 1-5 scales may seem in comparison to the reams of data collecting in other medical subspecialties, they’ll at least keep us in or near the ballpark and potentially head off common biases that commonly occur with pure guesstimation.
Too, their simplicity is nothing to scoff at the efficacy of simple checklists in medicine is a well-known phenomenon, and seeing that outpatient procedures are harder to collect data on in the first place[2], ease of response is more of a factor.
The best we could do would likely be to subject patients to some gauntlet of psychological evaluations both before and after surgery, not that the typical demographics of elective cosmetic procedures would be keen on enduring added hurdles. And that may just give us more noise than signal: I can’t get around the paywall, but the abstract of a paper titled “Plastic Surgery and Psychotherapy in the Treatment of 100 Psychologically Disturbed Patients” finds that “82% had a positive psychological outcome” and that “there were no lawsuits, suicides, or psychotic decompensations.” This is exactly in line with other studies of more typical patient outcomes, despite the fact that “Patients with psychological disturbances of a magnitude generally considered an ‘absolute contraindication’ for surgery” and were frequently rejected because of the results of their psychological evaluations.
But might we do better with a different angle of attack? With recent advancements in facial recognition and image processing technologies, perhaps a more quantitive assessment could be had. Jürgen Schmidhuber posits[3] that beauty, such as that of a face, is a function of regularity, or “compressibility.” The brain is stingy with its computing power, and thus has evolved to not only the ability to compress sensory data, but the intrinsic reward function to incentivize better ways of compressing data. How might the brain evaluate facial beauty in this model?
A neural network could conceivably be set up to either evaluate absolute compressibility, based on perfect symmetry, and before-and-after photographs of patients could be run through it to determine exactly how many bits the face lost relative to perfect compressibility. (In future version of this paper, I hope to have one or both of these built!) Some research has already been conducted on data sets of before-and-after pictures of patients.
Others[4] propose a U-shaped reward function for the brain’s compression algorithm—too regular, and the brain will lose interest; too random, and the brain won’t be able to arrive at any sort of compression at all. Somewhere in the middle lies the most beautiful, for they would have the highest ratio of apparent complexity to compressibility.
A temporal aspect, change in compressibility over time, might also be worth incorporating:
Schmidhuber’s theory is certainly simple and beautiful (for now-obvious reasons), but it’s a little “just so,” and doesn’t reference any neurological sources. Let’s go deeper.
II
The bottom line of James Scott’s Seeing Like a State is that (among other things) top-down systems sometimes work, and sometimes don’t. Allow me to make a superficial comparison: if you had to wager a guess, would you say the brain operates via a jumble of evolved processes, via metis? Or would you say the brain is a top-down autocrat imposing perfect rationality on the various sensors and sub-routines below it?
According to recent neuroscience research, the brain is unexpectedly, more like Le Corbusier and less like Ye Olde organically-evolved village! Predictive processing is something of a grand, unifying theory of the brain, states that two streams of data coexist in the brain—raw “bottom-up” data received from our senses, and a “top-down” stream of predictions based on our high-level knowledge of the world:
Famed neuroscientist Karl Friston elaborates on the primacy of prediction over perception:
Remember how I mentioned this is a grand unifying theory of the brain? It really can explain everything: dreaming, priming, the placebo effect, psychosomatic pain, autism, schizophrenia, phantom limbs, motor control. All of these are symptoms of a mismatch between what the top-down predictions of the brain are expecting to see, and what the bottom-up sensory data are actually seeing.
Even if the paper I referenced earlier claiming plastic surgeons operating on mentally ill patients have nothing to worry about, I doubt that any of them would touch one one subset of those patients: those afflicted with Body Dysmorphic Disorder (BDD). There are lots of studies that propose complex psychological screening frameworks for filtering out these patients. These tools are about as effective as you’d expect: 84% of plastic surgeons have unknowingly operated on a patient with BDD and, at least in patients with mild-to-moderate BDD, outcomes for this population are in line with the non-BDD stats (81% positive).[7][8]
But what if BDD is actually predictive processing error? Would that tell us anything about non-BDD patients who are unsatisfied with their outcomes? Would that give us more effective tools than post-hoc psychological narratives in shaping patient outcomes?
III
Cultural theorists often decry the “medicalization” of the body and the instantiation of a “clinical gaze” as a means of cultural control and oppression. That we give this gaze both the permission to deconstruct our bodies into isolated parts, and the sole authority to divine their singular hidden reality, reveals our implicit bias towards imposing arbitrary divisions that may or may not be accurate. Theorists correctly assert that rational systems like modern medicine work because we make them work, not because they are transparent conduits of eternal and absolute truth. By default, then, they engender our cultural ideals, nowhere more so than in cosmetic surgery.
Thus the rituals of the plastic surgeon’s office are suspect, because their medical functions might be auxiliary to their reinforcement of the system they are embedded within. Pre-operation consults are singled out, not as vital preparatory step, but as a type of conditioning:
The rationale for the procedure is also examined. That elective surgeries are performed within the greater medical paradigm of “do no harm” forces an attempt at reconciliation:
The dual nature of these practices shaping both outcomes and perceptions of outcomes is seized upon as a contradiction inherent in the very rationality modern medicine derives its authority from. Medicine’s privileged position of objectivity allows them only the former; to shape the latter is a betrayal of what medicine “is.” They’re pulling the wool over our blepharoplastied eyes.
But in the predictive processing model, these rituals, and even “medicalization” itself, might be the most powerful tools we have in affecting outcomes of cosmetic procedures. Instead of giving into the urge to develop more and more sophisticated psychological screenings and evaluations to get at “”what’s really there”“ in a patient, to learn “”the truth”” about an outcome only after-the-fact, what if we took predictive processing seriously and acknowledged that the gradient of clinical outcomes may flow the other direction?
Every book on negotiation tactics I’ve ever underscores the importance of getting people to buy into the negotiation process so that they’re more likely to trust the outcomes. What if each of these “rituals” of the plastic surgeon’s office, each touchpoint in a patient’s journey, from advertising, to before/after photos, to consults, to Photoshop-simulated outcomes, to payment, to the surgeon’s appearance, were studied and reformed with the knowledge that they play just as vital a role in patient outcomes as the physical results of surgery? I’ve found a smattering of research that gives a nod to this line of thought[11] (but even these still maintain that psychological screenings are most effective).
While this might be seen as a exercise in book-cooking, and real medicine doesn’t pull cheap swindles like that and may only use objective means to arrive at “”the truth,”″ consider that the empiricism modern medicine is based on is something of a swindle. Foucault’s thesis in The Birth of the Clinic[12] is that late 18th-century field of medicine was the site of an “epistemological rupture.” Rather than the incremental outcome of a plodding few centuries of accumulating data, the inauguration of modern medicine marked a sharp discontinuity in the field, rendering that which had come before utterly alien and incompatible with what followed:
This is not an isolated trend—legibility was hot stuff for the following few centuries. State planners couldn’t enlist architects with fetishes for straight lines fast enough to help them reboot intuitions onto rationality, onto the “absolute values of the visible.” Some domains stuck the landing better than others. But other than the replicability crisis, which is not a problem unique to the medical field, medicine has benefited tremendously from this shift away from metis and towards “objectivity.”
Even if it’s the case that doctors aren’t able to discern “reality” directly, and operate within an organization of knowledge that varies in its arbitrariness, and said arbitrarinesses can reinforce harmful cultural biases by default, and these biases often go uncontested because everyone assumes medicine to be the irreproachable paragon of rationality, no one would dispute medicine’s efficacy in spite of it lacking “ultimate” foundations and containing inherent contradictions.
I’m reminded of David Chapman’s summary of Robert Kegan’s concept of “meta-rationality”:
Post-modernist critiques like Foucault’s that warn us of inadequacies of building systems on rationality alone are entirely correct. But nihilism, faith in an ultimate lack of meaning, fails for the same reason faith in an ultimate meaning does: reality can stay complex longer than you can stay rationally enlightened. The fact that modern medicine is not based on some ultimate truth should not be the impetus to tear the whole thing down—quite the opposite!:
Sometimes, the theory is the program. Sometimes, treatment-resistant OCD can be cured by telling patients to put their hairdryers on the front seat of their cars.[14] And sometimes, you can knock on wood, cross you fingers, and move a bomb line over Bologna in the middle of the night.
Ramsey Alsarraf, Outcomes in Facial Plastic Surgery
Ibid.
Simon Brill et al, Anticipating Facial Transplant Psychological Issues
Jurgen Schmidhuber, Driven by Compression Progress
Nicholas Hudson, Musical Beauty and Information Compression: Complex to the ear by simple to the mind?
Karl Friston, Surfing Uncertainty
S. Higgins and A. Wysong, Cosmetic Surgery and Body Dysmorphic Disorder – An Update
V. Veer et al, Pre-Operative Considerations in Aesthetic Facial Surgery
V. Veer et al, Pre-Operative Considerations in Aesthetic Facial Surgery
Anne Balsamo, On the Cutting Edge: Cosmetic Surgery and the Technological Production of the Gendered Body
Ibid.
Michael Foucault, The Birth of the Clinic
David Champan, The Parable of the Pebbles
Scott Alexander, The Hair Dryer Incident