Tuesday, April 1, 2008

The Asymmetry of Uncertainty

Studying the human condition fascinates me. Above is an interesting clip from Nassim Taleb's recent talk at one of the Long Now Foundation's (click here) seminars on long-term thinking. I suspect that self-deception plays a role in understanding domain dependent skepticism, as Nassim has set out to study with Dan Goldstein at their Decision Research Lab (click here).

Robert Trivers (click here) of Rutgers University (click here) studies human behavior and self-deception. He views self-deception from an evolutionary biology perspective, and his seminal work has informed an entire school of thought. In one of his papers, "The elements of a scientific theory of self-deception," (found here) Trivers discusses some of the fascinating processes of deception that humans engage.

It turns out that the asymmetric emotional payoffs associated with uncertainty for positive and negative events tie closely with these self-deception concepts.

Thanks to collaboration with Navanit Arakeri (here), I read a fascinating study in the Journal of Neuroscience (click here to read), "Predictability modulates human brain response to reward." Researchers in this study conducted fMRI tests (learn here) on subjects to visualize activation of reward activity in their brains following consumption of small amounts of fruit juice or water in predictable and unpredictable ways. Juice and water were alternated and administrated at a fixed rate of 10 second intervals in the "predictable" experimental setup. In the "unpredictable" trials, the order of and the stimulant intervals for juice and water delivery were randomized. These random schedules were generated according to a Poisson distribution, producing a stimulant interval mean of 10 seconds. Before we discuss the results of this study, we must first note that Poisson distributions merge to Gaussian ("bell-curve") distributions for large values of Lambda (click here), so the underlying uncertainty tested in these "unpredictable" trials represents the "very mild uncertainty / unpredictability" of mediocristan (click here).

Here is a Poisson distribution. Notice that it converges to Gaussian by Lambda = 10.

In this study, Lamda = 300 seconds (5 minutes) / 10 seconds = 30, since thirty events occur per trial run, even in the "unpredictable" setup since these varied stimulant intervals averaged to 10 seconds. This is a statistical regress necessity that makes the math much more tractable, tidy, and easy to interpret and study. However, Lamda of 30 means that these researchers studied mild bell-shaped uncertainty, not the power-law uncertainty that we encounter in extremistan (read here) and in the wild as hunters. In addition, these researchers did not study impact (dosage size of fruit juice or water), which is a critical element to consider in power-law environments (impact trumps probability), so this study simply provides a platform for future inquiry. Nevertheless, these researchers found that "unpredictable" administration of fruit juice and water were associated with higher dopamine levels in subjects' brains. Since dopamine levels are associated with positive emotions, these researchers concluded that uncertainty could mediate our positive emotional responses: "The activity for rewarding stimuli in both the nucleus accumbens and medial orbitofrontal cortex was greatest when the stimuli were unpredictable. ... For pleasurable stimuli, these findings suggest that predictability modulates the response of human reward regions. .." Greater BOLD (Blood-Oxygenation-Level Dependent contrast; read here) responses were recorded in regions of the brain associated with reward pathways, which mediate responses to alcohol, drugs, sex, and other stimulants.

Given these findings, I developed some thoughts. Do drug addicts with easy access to drugs - in Los Angeles, for instance - achieve predictable, periodic drug intake, which then reduces the pleasurable feelings that these stimuli produce, which then creates a reinforcing cycle that drives addicts to increase impact (amount) continuously in order to achieve highs that were first/once experienced when drug usage was uncertain and unpredictable? Do these dynamics emerge with consumption of "energy" drinks like Throttle and Red-Bull? How about espresso consumption? Could we shunt habituation by evoking healthy self-deception to increase the uncertainty associated with consuming pleasurable stimulants? I suspect so. Moreover, I suspect that we could reduce habituation markedly by consuming bursty stimulants like espresso in power-law manners. Power-law uncertainty represents the Levy distribution of stimulants that we encounter in the real-world, and it gives us the best chance to keep our minds and bodies guessing about future consumption.

Here is a graph of a power law distribution (read here):

In my personal espresso consumption self-experiment, as noted in one of my previous posts, I have already determined that drizzling espresso on the foam of my non-fat latte to create a scarce and patchy (Levy, heterogenous) distribution of espresso in my drink enhances my emotional response (euphoria) to consumption - I hunt for the espresso within the foam and steamed milk. Diffuse (even, homogenous) espresso distribution, as encountered at Starbucks, reduces considerably the enjoyment that I experience when drinking a non-fat latte (the consumption process grows dull and repetitive; there is no diversity / variety from sip to sip). In my running experiment, I am testing the less tractable power-law uncertainty that the above neuroscience research scientists did not test. I will keep you posted ... I suspect this approach could help reduce habituation (I am also varying impact - espresso dosage - which could add another layer of uncertainty that could help prolong positive responses).

Next, here is something that I am passionate about: I hypothesize that sedentary humans living in the Information Age rarely push their physiologies into the far-from equilibrium states and processes associated with Power-Law training (we no longer hunt, and we no longer run from Tigers). Doing drugs, smoking, eating sweets, and drinking alcohol excessively represent ways to drive / cause transient spikes in our physiologies - eating a donut while drinking a soda certainly causes a spike in blood sugar levels. I suspect that addictions to these spike producing activities arise, in part, as outcroppings of inactive lifestyles that lack the physical and mental activities that historically, throughout our evolutionary histories, pushed our bodies into far-from equilibrium experiences (floods of dopamine release from playing an intense tennis match, for instance). Clearly, intense Power-Law training (click here) is missing from our lives and activity patterns in our current environments - we no longer hunt, gather, build shelters, nor protect our lives. I suspect that hyperstimulants like cocaine, methamphetamine, and nicotine appeal to and are addictive for so many people because their bursty effects mimic, at a shallow level, the physiological responses that our bodies used to (are built to) associate with power-law activities (fight or flight responses, for example). Leveraging this insight, I suspect that Evolutionary Fitness (here) and its accompanied Power-Law training, paleolithic diet (click here), and power-law sleep / lifestyle patterns could prevent habituation and could possibly cure detrimental drug addictions. By accepting that we are foreigners in the environments of the Information Age, we could construct a healthier ecology and replace drug-induced physiological spikes with physical and mental activity induced (via sports, as one possibility) power-law experiences that flood our blood streams with hormones that induce positive, euphoric, and health enhancing experiences: play some basketball; don't do crack (consuming crack induces Congestive Heart Failure, which leads to diseased sinusoidal, simplistic heart rhythms; playing basketball enhances and maintains multifractal complexity in our hearts).

Navanit also alerted me to another gem: a series of experiments published in the Journal of Personality and Social Psychology, "The pleasures of uncertainty: Prolonging positive moods in ways people do not anticipate" (read here). In these studies, Dan Gilbert of Harvard (click here) found that "people's positive moods lasted longer in the uncertain conditions. The results were consistent with the pleasure paradox, whereby the cognitive processes used to make sense of positive events reduce the pleasure people obtain from them. Forecasters seemed unaware of this paradox; they overwhelmingly preferred to be in the certain conditions and tended to predict that they would be in better moods in these conditions." The pleasure paradox captures an interesting phenomenon: we reduce our positive experiences associated with positive uncertainty when we explain and make sense of these experiences, yet we reduce our negative experiences associated with negative uncertainty when we make sense of and find meaning in these events. Basically, narrating, sense making, and meaning finding help us cope with, understand, and recover from negative events like tragedies (a car crash that kills a loved one, for instance). However, the paradox is that we reduce the happiness that we experience with positive events when we move to impose certainty on and make sense of and explain the uncertainties associated with these events ("I won the game because I practiced the most," as one possible explanation). In essence, Gilbert and his colleagues conducted a series of experiments (read about them in the linked paper above), and the results from these tests supported their pleasure paradox hypothesis, so they "argue that certainty can reduce the pleasure of positive events but that people tend not to recognize this in prospect and hence, under some circumstances, seek certainties that diminish their pleasure rather than uncertainties that prolong it." One way to benefit from this insight would be to leave the movie theater before a movie ends once you know that the movie will reach a happy ending - your emotional system will reward you with positive feelings for a longer period of time than if you find out exactly how the movie closes. Couples who wait until delivery to learn the sex of their child leverage this positive uncertainty dynamic as well. On the flip side, make sure to finish horror stories or murder mysteries because not knowing how these narratives turn out will punish your emotional system as your brain grapples with negative uncertainty.

Gilbert comments perceptively on the human condition: "Uncertainty is the cause of some of people's most debilitating anxieties, so it is not surprising that the human mind is designed to eradicate it. People gather facts, form opinions, and generate theories in an attempt to transform the unknown into the known - to make the world a bit less puzzling and more predictable by reducing their uncertainty about it." This statement captures the driving force behind Nassim Taleb's book, The Black Swan. The confirmation bias, the problem of silent evidence, and the narrative fallacy all originate from these cognitive dynamics that drive our thoughts, feelings, and actions. Living in a world that we don't understand pushes this drive for certainty, concreteness, and sense making into hyperdrive; alas, we are plagued by the availability heuristic: "There is little doubt that uncertainty is aversive much of the time, especially when an event is negative or when people do not know its valence - for example, whether the results of an HIV test will be positive or negative, or whether they will get a desired job or remain unemployed." Gilbert notes that people who write about traumatic events benefit from composing these narratives because our emotional systems reward us for making sense out of and bringing closure to the randomness and uncertainty of painful, negative experiences.

Nassim Taleb notes this asymmetry of uncertainty in his comments on the stock market's movements: many small gains, trumped by a few large losses. Our emotional systems punish us more for many small losses than they do for a few large losses, even when the overall loss amounts are the same, so it is no surprise that our cognitive functions and biases drive the gains / losses dynamics of stock markets since human interactions create stock markets in the first place. In fact, many social and cultural institutions represent manifestations of our cognitive processes and characteristics - self-affinity at work.

Thanks to the ever-insightful and supportive Dave Lull, I recently read two essays that provide medical examples and applications of the asymmetric emotional payoffs associated with uncertainty. In a recent article in The New York Times, "When the Disease Eludes a Diagnosis," Dr. Barron Lerner (click here to read; thanks to Dave Lull) discusses a patient who seeks certainty, understandably, in her diagnosis. Dr. Lerner's long-time patient suffers from a neurological problem that evades physicians' grasp: "While trying to be as sympathetic as possible, I find myself reminding Lucy of the limits of certainty in medicine. Despite enormous advances in technology, some diagnoses may remain elusive. ... But she remains convinced that she deserves to know exactly what she has. So we will continue to search." As the aforementioned research studies have shown, certainty certainly aids negative situations, but balancing certainty with uncertainty is a precious juggling act that astute physicians must engage - false certainty could be lethal and could close "the door to the unknown" diagnosis prematurely, the "What else could it be?" searching process. To impose false certainty on Lucy's condition and grab a diagnosis that does not fit her symptoms and experiences simply to give her a tangible, concrete label to assign to her condition would stunt efforts to unveil the true causes of her malady. Yet, as a sick patient, Lucy definitely stands correct to push her doctor to continue searching (her health is on the line); in response, Dr. Lerner elects to "continue the search." In reality, the search never ends ... especially when the Long Tail (read here) is involved (please read here and click here). My previously mentioned "overlapping barbell" strategies for medical judgment and decision making represent my efforts to "factor for the invisible" within the Long Tail of medicine.

Similarly, in another article published in The New York Times that comments on prayer and medicine, "Prayer, Faith and Doctors," (click here; thanks to Dave Lull) Dr. Robert Klitzman, author of When Doctors Become Patients (read here), writes about the important role that spirituality, faith, and religion play in patients' lives. Medicine cannot afford to shirk these forces because many patients cope with the uncertainties associated with illnesses by turning to religion, faith, and spirituality to find meaning and make sense of their negative, unpredictable experiences. Our brains and emotional systems long for the comfort that these three forces provide, so out of respect for our evolved cognitions and the dynamics of the asymmetry of uncertainty, doctors must integrate and respect the varied ways that patients bring certainty into, makes sense of, and narrate the wildly uncertain world of illness and disease: the human condition is a fragile one. To good health, my friends.

No comments:

Post a Comment