I’ve been thinking a lot about curiosity lately. Specifically, about curiosity in the sense of the proverb curiosity killed the cat: a potentially self-destructive pursuit of knowledge for its own sake that leads to unnecessary risk-taking. In humans such risk-taking often threatens not just the individual or even family/immediate group, but the whole species. Some people just have to go around figuring out new ways to blow things up, often with the noblest of intentions.
At a selfish gene level, the trait seems complicated, but not mysterious. The question that really interests me is this: how do our selfish genes fool us into being curious creatures, who sometimes get themselves killed, to teach our gene pools more about the environment? Altruism, a similar potentially self-destructive trait at the individual level, manifests subjectively as love (especially for kin), a sense of belonging to one’s community, and a capacity for attachment to some notion of greater purpose. What might be the analogous subjective experience for curiosity?
Curiosity does not seem to be a fundamental drive, unlike what I am told are the three basic biological drives (seeking pleasure, avoiding pain and conserving energy), so it is probably derived. Curiosity requires a certain energy surplus, since its visible signature is a restless dissipation of energy, but it does not seem directly motivated by energy conservation concerns. So is it derived from pleasure-seeking or pain-avoidance or some mix of the two? Does that make a difference?
I think it does, and I think the answer is that curiosity is primarily derived from pleasure seeking, not pain aversion. This has certain observable consequences.
***
Let’s get the genetic level of analysis out of the way first, so it does not distract us further.
You could view curiosity as simply the behavioral restlessness at the phenotype level that helps us break out of local optima on the fitness landscape at the genotype level, by introducing some randomness into the life-paths we trace through our environments.
Some curious cats get dead. Some get lucky. If a “curiosity gene” creates more luck than death overall, relative to populations that lack it, it will thrive. Individuals might pay the cost, as with certain kinds of altruism, but the selfish gene pool wins overall. All curious creatures either spread to occupy the planet, or kill themselves with their curiosity.
I doubt that there is such a thing as a “curiosity gene” but there is an idea I’ve mentioned before, called variability selection theory that sort of explains how we (and probably other archetypal curious animals such as foxes, cats and dolphins) grew larger, more programmable brains as an adaptation to uncertain environments that penalized over-adaptation to a single precarious niche.
That’s basically “curiosity” at the genetic level: the capacity of a gene pool to learn by turning noise into signal via wandering computers inside skulls, and using that knowledge to spread. Some such explanation, I imagine, will turn out to be correct.
So much for the selfish gene. I don’t really care about it much. Let’s get back to curiosity as a subjective experience of a psychological phenomenon in our individual lives.
***
For most of us in civilized and relatively safe environments, curiosity is clearly and rather obviously, pleasure-driven. We poke at something we don’t understand, usually with some sort of random access behavior. We hope to stir up something enjoyable. Stir the pot. Pot luck. Uncertainty and surprise are a necessary and inseparable part of the experience. The associated positive feeling seems to be joy. We will make fine distinctions among pleasure-words in a moment.
We can imagine a pain-avoidance motivation for a kind of “curiosity”: learn about the unknown so you can anticipate and avoid any potential pain lurking in the shadows. But this seems implausible to me. Pain aversion corresponds to fight-and-flight. Not to poke-and-pry.
The reason it is even on the table for discussion is that human organizations often seem to display curiosity-like behavior (exploring the unknown in open-ended ways) driven by pain-avoidance motives. Businesses and nations for instance, engage in institutionalized “research” not to feed curiosity, but to insure themselves against future uncertainties.
When humans do that — poking out things to fuel worry and pain avoidance rather than joy — we (correctly in my opinion) view that behavior as some sort of anxiety disorder. Morbid curiosity about potential disasters and worst-case scenarios, coupled with runaway survivalist tendencies, is a pathology. Curiosity about what makes the sky blue, or where that tunnel leads, or What Might Happen If I Push That Button, is more natural.
The difference isn’t trivial. Curiosity about potential disasters is anthropocentric and clearly related to a fear of mortality. It is not open-ended, but driven by relevance to survival concerns. It is problem-solving masquerading as curiosity.
Curiosity as an open-ended seeking out of joy in the unknown seems to go the other way: taking on mortality risks in order to experience joy, with or without material benefits.
It may be pragmatic to anticipate pain and prepare for it (by buying insurance, hedging investments or getting inoculations for example) when the costs are sufficiently low, but conceptually linking curiosity to worry seems to be an unnatural thing to do. For one thing, the experience of this sort of anticipatory thinking and contingency planning does not seem to be the same as the experience of being caught up in a curiosity-driven adventure.
Insurance is most successful for all when nothing happens. Adventure fails for all when nothing happens.
***
As I already noted, energy conservation seems like an implausible source of curiosity, but may have a different relationship to it.
We instinctively associate idleness with curiosity. It might even make sense to define idleness as the state of having more energy than is needed for the pursuit of immediate ends. In such a state, humans more naturally dissipate than conserve. Saving is a learned behavior, waste is natural.
Mathematicians and engineers sometimes say they try to be creatively lazy by finding shortcuts or formulas that automate tedious tasks. But a little thought immediately suggests that is not energy-saving behavior or efficiency seeking. This is boredom-avoidance.
The fruits of boredom avoidance may be cashed out in the form of increased efficiency, but energy efficiency is not what motivates it.
So what is boredom avoidance?
Boredom is partly a kind of pain, but I’ve concluded that it is more than that. Boredom is also natural curiosity curbed. To view it only as pain is reductive. The natural state of a living human might not be a state of restful equanimity. It might be a state of casual ongoing exploration of the unknown.
This is reflected in the proverb that the idle mind is the devil’s workshop. A mathematician or engineer who figures out a labor-saving calculation, device or program will generally proceed to waste the saved energy by inventing further complications to keep himself/herself amused.
If it ain’t broke, it just doesn’t have enough features yet.
***
We’ve complicated the picture quite a bit now. It seems like it would be useful at this stage to more carefully distinguish between three different varieties of “pleasure.”
I am going to restrict the word pleasure itself from here on out to predictable, hedonic, sensory pleasure, of the sort that quickly hits diminishing marginal returns via hedonic adaptation. A marker of hedonic pleasures is that other humans are often involved in supplying them, which means hedonism is strongly related to power.
Two other varieties we can distinguish are happiness, which I will associate with non-exploitative relationships (happiness in the sense of Jonathan Haidt), and joy, which I will associate with curiosity and its fruit, meaning (as in “the joy of discovery”). In English usage, the three terms often bleed into each other, but let’s keep them separate here even if it feels somewhat artificial.
These distinctions are not entirely arbitrary. Pleasure, happiness and joy are the varieties of positive experience associated with the three basic positive-action drives of social species respectively: getting ahead, getting along and getting away. The first two are derived from human attachments, while the third is a relatively detached mode of positive experience.
If you want pleasure, you seek power. The absence of pleasure is hedonic deprivation.
If you want happiness, you seek relationships. The absence of happiness is loneliness.
If you want joy, you seek mystery: The absence of joy is a deep sense of existential meaninglessness.
The opposite of the pleasure of winning is the ignominy of defeat.
The opposite of the happiness of a relationship is the sadness of a loss.
The opposite of joy is fear.
You choose in pairs, or not at all.
***
Getting ahead is winning: a concept that has meaning mostly within a social context involving members of the same species. Not just the ritualized kinds of human SAT-score and medals victory, but things like the alpha lion getting to eat the choicest bits of the prey first, the pick of females, and the nicest spot on the grass to sprawl out on. Getting ahead — and therefore pleasure — is biologically the most basic, since it directly couples individual success with genetic reproductive success.
Getting along — and therefore love — is not as basic, and has a weaker link to genetic success. All species “seek pleasure” in some sense, but only some are social enough to seek happiness through emotional relationships that extend beyond maternal ones. You participate in a broader way in the success of an entire related gene pool. Happiness seeking is also about us-versus-them distinctions. To relate to some deeply, it appears you generally have to detest others equally deeply. Love without hatred is political fiction.
Getting away — and the joy of discovery — is the furthest removed from biology. It seems reasonable to say, in English, that we enjoy an unexpectedly beautiful sunset, but there is no hedonic pleasure in the sense of winning a lion’s share of a scarce good through competitive success. There is also no love or belonging (or hatred or estrangement) in the sense of human relationships.
Genetically, it makes sense that curiosity would lead to the most detached, individualistic (and volatile) form of positive experience. The material fruits of curiosity are not just rare, they generally accrue to gene pools far away in time, space and genetic distance. There is no clear correlation between being interested in astronomy and number theory today and your genes or your third cousin’s genes being more successful. You might even discover things through curiosity that primarily benefit zebras in the year 2405. Worst of all, you might discover things that kill you.
By contrast, the material fruits of getting ahead and getting along accrue more locally in time, physical space and gene-pool space.
And since curiosity can in fact kill the cat, it also makes psychological sense that it is not as anthropocentric a drive as the other two.
***
You can also distinguish among the three types of positive experience in terms of the role of uncertainty in each.
For getting ahead, uncertainty is mainly about outcomes. Especially competitive outcomes. For getting along, uncertainty is mainly about trust in others. For both, uncertainty is something to be eliminated. We prefer certain victories to uncertain ones. We prefer faithful friends to unreliable ones.
But with curiosity, uncertainty in the path is actually what drives the enjoyment derived from the behavior. Removing the uncertainty impoverishes the experience. But curiosity is more than mere variety-seeking, which is a self-destructive trait. It is about variety-seeking with an eye to discovering meaning.
When curiosity contaminates competitive instincts, we often give up a winning edge to challenge ourselves more and learn more about the unknown parts of ourselves. When curiosity contaminates relationships, we try to introduce variety, see new mysteries in comfortable relationships, or seek out unpredictable and tumultuous partnerships.
But curiosity can also find expression in pure forms. And so from time to time, we pack our bags, buy tickets, and head off in random directions, both inward and outward, to potentially unpleasant and unhappy places. And we still enjoy ourselves.
Related to uncertainty is the idea of humor. Humor seems to always involve some sort of surprise, but not all varieties of surprise are the same.
Getting ahead humor usually involves the ups and downs of status shifts, and is often cruel. A court jester may occasionally sass the king, but he is more likely to borrow the king’s power and make fun of someone weaker.
Getting along humor is usually in-group humor at the expense of out groups. It too is often cruel.
But getting away humor, almost by definition is about laughing at yourself, because there is no one else around. It is chagrin-and-grin without an audience.
***
At the risk of death by pun, I will call the subset of individuals for whom curiosity is the dominant drive in their personalities the Joy Luck Club.
As in that story, joy and luck are inextricably linked (in the story by the mah jong motif, in this essay by curiosity). Seeking joy through engagement of true randomness — not just the good bits — is a form of feeling alive through play. This is not a case of trying to be clever enough to win in the presence of the unknown or trying to position yourself to enjoy the upside of success while avoiding the downsides of failure (yes, we’ll get to Taleb in a bit).
That’s a winner mentality. That is about skill and intelligence deployed in service of getting ahead, outliving and outlasting others, and keeping your own mortality at bay as long as possible.
Curiosity is not driven by a winner mentality. It is not about skill or intelligence, at least not primarily. It is not about preserving life against the forces of mortality. Play is about engaging mystery to possibly derive new joy through surprises, chagrin, humor, discoveries and inventions. The cost, outside of controlled play environments for children, is risk of death. Not just for the individual, but beyond.
There is a certain sense in which skilled games like blackjack and poker are less fun than unskilled ones like playing the slots. When skill is cordoned off from fate, what we experience is pure curiosity about our uncertain future.
I believe this trade-off is fundamental. To seek immortality is to be joyless in a certain sense. You may enjoy a certain amount of getting ahead and getting along — accumulate billions and live to see your great-great-great grandchildren perhaps — but to experience joy, you must fundamentally pursue luck without much intelligence, and take on mortality risks.
Curiosity may have killed the cat, but a lack of curiosity is certain to kill joy. Incurious people are usually the biggest killjoys around.
I am dead-curious about this.
***
We part ways with foxes, cats, dolphins, magpies and parrots when it comes to what we do with the fruits of curiosity, and how that drives our further curiosity.
Alone among the curious animals (though this seems like a conceit that more research might invalidate), we seem to be curious about clearly useless things. Or at least, things that have no obvious and immediate use. Humans seem to frequently poke at things that yield returns, if at all, only generations later. And often in ways unsuspected by those who do the poking.
We stare at the stars, we peer through microscopes, we climb mountains and we dive to the ocean floor.
This behavior, so natural to humans, is incomprehensible to human organizations. So things like space programs or other pure curiosity driven efforts have to be justified by politicians on the basis of “will improve life here on earth through the discovery of new materials and advances in medicine.” This is probably the mother of all idiotic fictions. Fortunately, we don’t seem to require our institutional fictions to be credible. Merely sufficient to stop conversations we don’t want to have.
There is an interesting symmetry here. Organizations naturally try to avoid pain — the pain of business model obsolescence or national decline for instance — through institutionalized “curiosity.” They find joy-seeking unnatural and in need of justification (hence the paradoxical notions of “efficient” innovation with high “yield” or “impact” and the relentless war on waste).
This has even been turned into a depressingly banal formula for innovation: what pain are you seeking to relieve?
For humans the reverse is true. Curiosity driven by pain-aversion is unnatural, but curiosity driven by joy-seeking is natural and requires no further explanation. Efficiency is the last thing on our minds when we are being curious. The concept does not even apply: efficiency pre-supposes a goal. Waste is pain in the efficient pursuit of goals.
***
This situation makes complete sense once we recognize that we still haven’t managed to create organizations that are fully living in any sense of the term. Organizations share our capacities for pain-aversion and energy conservation, but not our capacity for curiosity. That means, when they aren’t serving the needs of individual powerful (sociopath) humans, they are primarily driven by self-preservation instincts.
It is possible to codify pain avoidance into organizational forms relatively easily. It is also easy to wire organizational foresight capabilities to pain-avoidance. Contracts, incorporation models and insurance policies codify pain-aversion instincts. Corporate law is a product of this engineering. “Innovation” is mostly business model insurance. When it manages to be more, it does so mostly by accident.
Pleasure and happiness sort of have weak cognates in the language of corporations. Pleasure for a corporation is winning a market and living longer. Happiness for a corporation is harmonious relationships with supply chain partners of the sort the Japanese call keiretsu. I suppose you could say that M&A activity is a sort of sex for corporations.
But there is no such thing as a curiosity-driven joy-seeking corporation in my experience. Individuals within corporations may be driven by a stealthy curiosity (or a tolerated 20% curiosity), but corporations as a whole don’t seem to be curious in any recognizable sense of the term.
I would like to invent a model for the Curious Corporation™ but I’ll leave that for another day when I am in more of a consultant mood. For now, I’ll just stake my claim to the term and elaborate in the unlikely event that someone pays me to.
Let us just say for now that by extrapolation from human beings whose curiosity is primarily morbid and neurotic, and focused exclusively on anticipating and avoiding risks, all large organizations are joyless wonders driven by morbid and neurotic curiosity at best, seeking to live for ever, and capable only of seeking market-winning hedonic pleasure and a certain amount of keiretsu and M&A happiness.
No corporation ever climbed a mountain just because it was there, or used a strong cash position to build the corporate equivalent of a Hubble Space Telescope. I would be delighted to learn of any counterexamples (be careful while considering this question though: we are talking of corporations themselves, not of human curiosity expressed through corporate leverage by individual powerful people).
***
Let’s consider the question of curiosity from the point of view of its absence: the state of joyless, pleasurable happiness (Julian Barnes has an interesting fictional exploration of the idea in A History of the World in 10 1/2 Chapters).
I only recently realized that for a large number of people, including (surprisingly) very knowledgeable and intelligent people, knowledge is not a source of pleasure. An example appears to be Nassim Nicholas Taleb. Despite his erudition and knowledge, and voracious reading and learning, he does not seem to actually enjoy knowledge. He claims in Antifragile that he would rather be happy in a world he does not understand than miserable in a world he does.
That this is a false choice is the least of the problems here. The bigger problem is that such a view gives up the joys of curiosity. It almost sounds like a view of knowledge as an unpleasant necessary burden, to be eliminated where possible.
The view suggests that the pleasure and happiness of getting ahead and getting along on one’s own terms, of redemption of a social variety, is all you need. I do not think this is true. You can live a pleasurable and happy but joyless life — one lacking the fruits of curiosity — and you will feel that something is missing. We are more deeply wired for restlessness and trouble-making than we think. We want to know. We want to understand.
But there are many people for whom curiosity is not a a powerful enough drive to make a difference. I suspect they are the majority. The Bilbo Baggins types, easily tempted into adventure, are not very common.
That is why novelty and innovation have historically been viewed with suspicion — a suspicion Taleb, among others, hopes to revive. But I’ll leave a full review of Antifragile for later, after I’ve had time to process that complicated book. For now, I’ll just say it is a worthwhile book. Very thought provoking once you get past all the tedious anticipatory trolling and baiting of critics.
***
In terms of the distinction between appreciative and manipulative (or instrumental) knowledge that we’ve talked about before, you could say that a joyless approach to knowledge is an approach that seeks out only the manipulative aspect of knowledge, and ignores, or worse, actively resists, the appreciative aspect.
The realization that there are people who only relate to knowledge this way has sort of rocked my world. Besides Taleb, there are plenty of others who operate this way. Many machine learning experts and champions of phenomenological domain knowledge think this way, as do the sophisticated defenders of religion for its instrumental effectiveness. The argument goes that non-expert heuristics defeat expert models, and that naive science does worse than religion in making people happy.
True, but besides the point. Expert models and naive science exist in the human world because humans have an independent desire for larger, deeper meanings, and find joy in appreciative knowledge that does not require justification through a demonstration of instrumental effectiveness or even validity. Today the Big Bang theory is considered true and Fred Hoyle’s steady state model is considered false. But both are sources of appreciative satisfaction.
Our quest for meaning has its ups and downs (in fact it is those ups and downs that constitute the rewards of curiosity). Sometimes we exhibit remarkable naivete and let our enthusiasm for appreciative joy lead us astray. We make up bullshit theories and authoritarian high-modernist architectures.
Other times, we make progress, as in the case of moving from creationism to evolution.
But these considerations are irrelevant if the satisfaction of curiosity is the primary drive.
***
For the curious, it is the joy of meaning that is central to the search for meaning, not the practicality or utility of meaning. And that joy lies in the volatility that is inherent in its pursuit. This is something that Taleb, oddly enough, seems to miss. In his ambitious study of things that gain from disorder — a getting-ahead/winner concern — he doesn’t pay much attention to things that might enjoy disorder, independently of whether they gain or lose from it. Of course, this criterion only applies to things capable of subjective experience.
To return to curiosity, if we agree that it is a relatively detached form of pleasure seeking, and if we agree that a good deal of what we figure out has no immediate practical value, we must conclude that seeking appreciative knowledge is not primarily a utilitarian behavior at the level of a single individual. In fact, the voluntary seeking-out of potentially mortal dangers, without clear a priori proof that they can be surmounted, is part of curiosity. At least in the wild.
So basically, curiosity is perverse. Russian roulette is no fun without a gun.
This is surprising to some people. They are surprised to find out that learning and exploration are not just transient costs to be recouped later after “winning” a race of some sort. That there are people who get bored and leave just as the extrinsic rewards start to come in. That these are in fact the truly curious people who engage in open-ended exploration in the first place. That people who compute the upfront costs and future benefits of curiosity generally conclude that it is not a worthwhile behavior.
This idea that curiosity is its own reward, worth giving up other things for, is captured neatly in the Garden of Eden myth. It is the fruit of the tree of knowledge of good and evil that causes the fall of Adam and Eve. So why did they choose it?
They were curious.
My own experience with curiosity validates this intuition. It was simply enjoyable for me to learn, as a kid, how stars were born, and how they matured and exploded into supernovae, leaving white dwarfs behind. It was simply enjoyable to explore alleys I hadn’t been down before and figure out how long I could hold my breath.
Play is fun, play can be destructive.
Eventually, this seeking out of appreciative knowledge for pleasure did bring some modest rewards, but that is not why I pursued it. There are far better paths to rewards than curiosity.
***
There is a movement afoot today to devalue the very idea of meaning as in meaning of life. This philosophy can be stated simply as follows: everything that makes life worth living derives from privileged sensory experiences derived through instrumental action: doerism.
Doerism holds that knowledge ought to be viewed in purely utilitarian terms, that manipulative knowledge is more valuable than appreciative. That efficient knowledge is better than discursive knowledge. Knowledge that is both manipulative and amenable to being acquired and applied by a machine learning algorithm, without human participation, is superior to knowledge where human aesthetics and judgment must clumsily intervene.
Perhaps the most insidious idea in this movement is this one: understanding is not necessary for action and is therefore unnecessary and even dangerous.
This idea is insidious because it is completely true, but incomplete in a way that makes it irrelevant.
Yes, appreciative knowledge is often unnecessary for reaping the rewards of action (I am even happy to concede, though this is not actually true, that it is entirely unnecessary and that all manipulative knowledge can be robustly derived from phenomenological understandings of local domains through trial and error by our doers).
It is also obvious that appreciative knowledge pursued for its own sake can be a dangerous thing, and that impoverished appreciative knowledge applied as a design aesthetic can lead to tragedy (this is authoritarian high modernism, which we’ve discussed often).
So the antimeaning movement is really a restatement of the Garden of Eden position: do not eat the fruit from the tree of knowledge of good and evil.
The recent scare around the Higgs boson is a modern Garden of Eden story. I have no idea what the Higgs boson is, despite a recent visit to CERN, where I got an opportunity to tour the facilities and play a spot-the-Higgs-boson video game. But whether or not it is true that finding the Higgs boson might destroy space-time (it seems to be a popular misunderstanding), the fact that some people view that concern as sufficient reason to not look for it is very suggestive. Active avoidance of potentially dangerous knowledge — call it anticuriosity — is a deep-rooted cultural tendency, even if it is not a very natural human one.
So there are two types of people in the world. Those who say, “Higgs boson, damn the Universe!” and those who say “No!” to the idea, and other related Eve’s Apple types of knowledge (creating artificial life in the laboratory, gene-hacking, the truth about aliens).
***
Meaning, quite simply, is the joy extracted from truth. Not any kind of truth, but a very specific kind: truth as captured in appreciative models. Reach-exceeds-grasp models.
For the gonzo experientialist, who would blow up the universe for a Higgs boson, joy is to be found in truth through lived meaning. Psychologically, models are appreciative first — sources of pleasure — and instrumental second (if at all).
It is not necessary to have a theory about how appreciative models deliver pleasure to biological sacks of entrails, but since you and I don’t need a reason (let alone a practical one) for theorizing, here’s one that sort of amuses me: appreciative models are truth-pleasure prosthetics.
What do I mean by that?
Pleasure derived directly from the senses — a capacity we share with other creatures — requires no symbolic mediation. Pleasure and happiness — the getting ahead and getting along varieties — are both ultimately sensory.
Appreciative joy derived from symbolic models on the other hand, requires more work. It is still ultimately derived from sensory experience, but it is the added symbolic meaning that makes it worthwhile. A bag of chips needs no theory to be enjoyable. Neither does the company of a friend. But without our capacity for reading meaning into things, much of the pleasure in the universe remains inaccessible to us.
This, ultimately, is the poverty of doerism. It allows human vanity and narcissism to limit human experiences. To say that the realities we cannot skillfully manipulate are not worth knowing about or exploring seems to be a joyless way to approach life.
Those who lack the capacity to appreciate useless meanings tend to fall asleep in planetariums.
As Robert Browning once said, Man’s reach should exceed his grasp, else what’s a heaven for?
***
While I am sympathetic to the concerns of the anticurious, I am fundamentally about voting Yes to curiosity. Not because I am immune to the known pleasures of the senses that the anti-meaning types treasure, but because I am not blind to the appreciative pleasures of entirely useless and potentially universe-destroying knowledge.
A bird in hand is not always worth two in the bush to the curious.
If the devil offered me a deal where there was a 99% probability I’d be told the secret of wormholes for hyperspatial travel, and a 1% probability that the universe would be destroyed, I’d immediately take it. I wouldn’t even consult the rest of humanity or call for a vote.
The interesting thing about the existence of knowledge sociopaths like me is that human security-seeking is doomed, not just because of outright criminals on Wall Street exposing the rest of us to Black Swan disasters but because a non-trivial portion of humanity thinks like I do, not out of greed or a desire to exploit others, but out of simple curiosity and meaning-seeking.
We are Pandora. All your fates are belong to us. Because we cannot resist asking the questions whose answers we probably don’t want to know and messing with and unleashing forces we probably cannot control. Curiosity is unfortunately not a democracy-friendly trait.
The unreliable coupling between seeking meaning and being more effective or happy in the real world does not really hold us back because we do not seek meaning to be more effective or happier in the real world.
That’s a gamble that may or may not pay off for some random person or zebra in the year 2533. Curiosity, in this lifetime, for you and me, is not about winning pleasure or finding happiness in love.
It is about prodding the unknown with a stick for the hell of it and battling ancient existential boredom with new meanings.
If this means stirring up unnecessary trouble, so be it.
***
I haven’t said much about immortality or longevity, but it is an idea that hangs in the air whenever you think about curiosity. It’s been part of the subtext through this whole piece.
Here’s an angels-on-a-pinhead question. Are immortality-seekers necessarily joyless?
I think they are. If you cannot die, variety is not meaningful.
On average, curious agents probably die more quickly and in greater numbers, than incurious ones. This means the basic trade-off in human life might be between longevity and joy (which I have reductively defined as the sort of positive experience one gets out of yielding to curiosity).
In a way curiosity in humans is basically a death wish of sorts. A desire to at least control the course of mortality, if not its ultimate destination.
But this trade-off may not exist for organizations. I suspect that the way to create immortal corporations is to first try and create curious corporations.
If that turns out to be true, it would be very ironic.
I think a lot of the people you call “immortality-seekers” might be more fairly classed as “against loss of agency from ongoing aging”. I can’t see a fundamental conflict in living curiously for 300 years before getting blown up in a Martian lander accident.
Looking forward to the antifragility post, Taleb’s been echoing round in my head lately too.
Your “[curiosity] is about variety-seeking with an eye to discovering meaning” seems very similar to Schmidhuber’s description of curiosity, which you’ve probably heard of, as the search for information that best improves compression algorithms over streams of sensory input. I expect your use of “meaning” is somewhat different from Schmidhuber’s, which shouldn’t be surprising since he’s one of those machine learning experts.
But I also suspect that human joy over acquiring useless knowledge is less about selfish genes furthering the species (group selection is pretty hard to establish, after all) and more about a standard value-of-information calculation: any given piece of knowledge is probably useless, but has a small probability of a large payout and usually costs little to acquire (though there are indeed risks). Our joy in gaining knowledge is indeed intrinsic, but that doesn’t prevent knowledge gained in this way from being useful and helping our genes propagate — in fact, you’d expect a net-positive behavior (for the organism) to feel intrinsically good.
No, I am not familiar with Schmidhuber. The connection to compression algorithms makes sense if you’re talking Komogorov, not so much if you’re talking Shannon.
And yes, I think I mean something different by meaning: mine is a relative notion. The most compact subjectively comprehensible account of something relative to the cultural context of an agent. Or something like that. Information theoretic models of meaning don’t usually compute relative information content or code for comprehensibility from a given subjective perspective (which is why the efficient solution to a Rubik’s cube position by a machine algorithm looks crazy to human expert solvers for instance…)
I think we do differ on the utility speculations. I would think most rewards of curiosity are quite unlikely to accrue to the person who discovers the valuable new thing. There is some anecdotal evidence of this. For example, pioneering companies rarely own the markets they create (they average about 7%). Curiosity kills market pioneers.
I would suggest, as TL;DR:
“If it ain’t broke, it just doesn’t have enough features yet.”
There is a risk to finding out new things.
But there is often a greater risk in staying in your comfort zone.
Creatures who explored new ways would have found ways out of dead ends and inevitable death situations such as hunger. Curiosity would have resulted in more food from exploring “just around the corner” or “wonder if this works” ways of hunting. It also drives collaboration as one has the idea then others refine it into something which works repeatedly – this gives higher status to the ideas person and higher status gives greater chance of survival.
There is a misunderstanding of Darwinism here.
It is not that behaviour is rewarded and thus a feature develops.
It is that a feature develops and runs parallel with the existing.
An event then occurs which changes the situation and this favours one or the other.
We have millions of these traits running in us at this very moment – just waiting for the event which favours it and makes it life critical.
I don’t think that Taleb has an issue with curiosity or with innovation. It might seem so at first because he has a distaste for things that are overly-theoretical; but the first thing I thought (as someone who has read him nearly religiously–if you excuse the play on words) about what you were saying was “makes perfect sense, if I take a risk as an entrepreneur/inventor, the upside is a new technology for everybody, the downside is the financial ruin of a single human being out of billions.”
But I nonetheless see what you’re saying about “doer-ism”, and it caused me to not care much for startup culture–too much emphasis on the idea that everything is pain-relief, as you put it.
I also don’t think it’s untrue that space-programs and other such speculative ventures are worthless from a pragmatic standpoint. One of the messages of antifragile is that we simply don’t know what’s going to happen next, so it’s worth our time to seek out unknown-unknowns. It doesn’t matter what going to the moon did or didn’t accomplish for us–it was a novel enough enterprise that it would be foolish not to make the investment.
To make an analogy with Taleb’s simple philosophy of investment: curiosity-seeking is the high-risk/low-probability part of the portfolio. Getting ahead and getting along are more like cash (if I wanted to overdo it, I’d say that “getting along” is cash and “getting ahead” is something more middling “blue chip” stuff [read: seeking validation has subtle but serious risks]–but I don’t see the need.)
There is one area where I disagree with Taleb, however; but it’s funny because it’s based on its own logic. It actually reconciles your own disagreements with him about appreciative models. Appreciative models are the source of cheap tricks. Why do I say this? Consider what an appreciative model is: it’s an informational compression of our observations. If you have a very simple model that explains a lot of complex phenomena with sufficient accuracy, then those principles can be leveraged to create a strategy that is lower-cost in some way, shape or form.
But one more condition, one that is consistent with the logic of Tempo: theories, as Taleb himself put it, are fragile. They “platonify” (and I hypothesize that mathematically, they always replace E[F[X]] with f[E[X]]). This is (at least part of) the essence of why cheap tricks are eventually destroyed by entropy. But extending on his logic of optionality, what matters is that we have the option of leveraging the cheap trick.
Ultimately, what makes an appreciative model what it is is that it’s a narrative. Narratives can be abused, especially when they’re mistaken for proof, but they are also the generator of epistemic leverage.
In Boydian terms: We need appreciative models in order to build snowmobiles.
Also, regarding machine-learning:
By my analysis, machine learning is a case of naive “doer-ism” (I think you just mean “pragmatism”.) On a mathematical level, our reality is fractal, and I think that what that suggests is that we need to think in complex layers to find patterns. Traditional machine learning algorithms are fundamentally impoverished because they’re trying to infer patterns from a completely static and flat numerarie.
I think that my point is empirically supported by the fact that a ~100,000 fold increase in computing power over the past few decades has clearly not produced anything even remotely close to a ~100,000 fold increase in scientific advancement.
> By my analysis, machine learning is a case of naive “doer-ism” (I think you just mean “pragmatism”.) On a mathematical level, our reality is fractal, and I think that what that suggests is that we need to think in complex layers to find patterns. Traditional machine learning algorithms are fundamentally impoverished because they’re trying to infer patterns from a completely static and flat numerarie.
I doubt any machine learning researcher would recognize your claims or descriptions; there are plenty of hierarchical algorithms and strategies in machine learning – to name the most famous example, deep belief networks.
Interesting. Is there an ML technique you’d say deserves to be called a “curiosity driven algorithm”?
BNs/DBNs don’t seem that way to me. They’re purposeful. They may learn in inscrutable ways that humans can’t parse, but I don’t think they try to learn “useless” things, do they?
Venkat, I think you’re missing a fundamental point about what is “useless”. “Useless” knowledge is arguably the equivalent of “junk DNA” or “spare parts”–they’re not immediately applicable, but they may turn out useful in later situations.
An algorithm could do this, at least to an extent, but the approach of “machine learning” doesn’t seem to tackle this (not from my experience, at least.)
> Is there an ML technique you’d say deserves to be called a “curiosity driven algorithm”?
In the simplest possible sense of ‘curiosity’, I’d say there’s plenty of algorithms which place value on estimating the ‘value of information’ and acquiring additional data and refining estimates: ‘multi-armed bandit algorithms’ come to mind as a classic approach to the question of how many resources or pulls of the lever to spend learning about the various payoffs vs doing the current estimated-optimal thing. It’s not just applicable to optimizing webpages, that sort of ‘sequential testing’ approach was pioneered for quality control and then clinical trials.
Less simply, Schmidhuber has an interesting conception of curiosity and beauty as maximizing the second derivative of the compression of observed data, IIRC, which I think matches up in some ways with your post. Check out http://www.idsia.ch/~juergen/interest.html and http://www.idsia.ch/~juergen/creativity.html
gwern, I paused before responding to give what you said some thought. I think that I did jump the gun and I should reduce my assertion to a suspicion.
My experience with machine learning came from learning about the basic prototypes for things like classification trees, neural networks, and expert systems; so it’s probably unfair of me to criticize machine learning on those grounds. Nonetheless, it does seem to me that adding hierarchy to the algorithm, while certainly letting it deal with things in a more complex manner, still has to distill everything to a numerarie that it can process. Neural networks offer a bit more promise, but I highly doubt that more nodes and links is going to massively improve computational cognition.
This on its own isn’t a problem, because one could tailor these networks to deal with a specified domain, but the issue is that it does seem that machine learning algorithms are treated as some kind of Rosetta Stone that can extract all the relevant patterns out of any domain:
http://stdout.be/2011/07/18/machine-learning-fairy-dust/
“the capacity of a gene pool to learn by turning signal into noise via wandering computers inside skulls”
I think you meant “noise into signal”. No creature never got nowhere turning signal into noise.
Well, technically cognition actually produces a net surplus of noise. The negentropy is local/temporary (Maxwell’s Demon).
> No creature never got nowhere turning signal into noise.
‘Cryptography’. ‘Mixed strategies’. ‘Probabilistic search’. ‘Randomized experiments’. Ringing any bells?
Fixed, thanks.
Alexander and gwern: I see your point, but this really was a typo :)
This post helped me tie a few ideas that have been running around in my head.
Your “getting ahead, getting along, getting away” idea maps nicely to something I myself believe in, which is that we, as human beings, inhabit 3 worlds simultaneously – the person himself (mind and body), human society, and the real or natural world. (Also, I tend to think of your “getting ahead… etc.” as not so much a sequence of phases in life, but more of an “asset allocation” i.e., depending upon your situation, you may allocate 70% to personal world, 20% to human society, and 10% to the natural world, or some other combination.)
So, I guess “getting ahead” would map to increasing the chances of survival of the personal world. “Getting along” would map to helping human society improve its chances. And “getting away” could be though of as mapping to helping the natural world overall. After reading this post, I realize that may be I can extend this mapping further: “pleasure”, as you have defined it, would map to positive experiences in the personal world; “happiness” would map to that in human society; and “joy” could map to creating a positive experience / outcome for the natural world overall. This may help to explain the existence of curiosity as necessary for increasing the chances of survival (or at least positive experiences while alive) for the whole of the natural world. This may also help to tie curiosity back to the biological imperative i.e. in order for the person to survive and have positive experiences, the human society, and beyond that, the natural world, needs to survive and have positive experiences in the long run.
I think you’ve mischaracterized Nassim Nicholas Taleb’s position (whom you incorrectly called “Nicholas Nassim Taleb” — I thought you might want to fix the typo). Anyway, you wrote: “He claims in Antifragile that he would rather be happy in a world he does not understand than miserable in a world he does.” I’ve only read the book once, but I don’t remember ever addressing that dichotomy. Rather, he claims that he wants to live happily in a world he doesn’t understand.
I think he starts with the Kantian assumption that we live in a world that we not only don’t understand, but can’t understand due to the fact that we can never shed our human spectacles and see things as they really are. If you accept that assumption, which I do, then it’s easier to understand his philosophy.
I don’t want to put words in his mouth, but I think he’d agree with me that curiosity is fine, so long as it doesn’t involve a transfer of fragility to others. Also, I suspect that too much curiosity will make you fragile.
I don’t think Taleb has a consistent position intellectually, or wants one. He has consistent values though, which creates a different kind of coherence.
I’ll leave this for later. Will correct the name, thanks.
I agree with your point about Taleb. My major contention with Antifragile (which I wrote about in my review of the book) is that it seems contradictory to be both a humanist and a proponent of antifragility, both of which Taleb claims to be.
When it comes to antifragility, it depends on which level you’re looking at things from. We want to harness the antifragility in some systems without realizing that something just may want to harness the antifragility in us.
Yeah, that was my basic conclusion too. The humanist-antifragility thing. Ties in to your old guest post about immortality-seeking (or longevity seeking) in general.
There is more down that bunny trail than has been properly explored I think.
You’re a genius. I’ve studied curiosity as a vital component of a human-level artificial intelligence. Infants are deadly-serious when they play. If you’ve ever tried interrupting a baby just as they’ve discovered a new schema (in the Piagetian sense), they can get furious. Discovering new reliable bits of knowledge about the world is the most important thing humans do, period. And they are hardwired to get a rush of pleasure from it that is as fundamental as sex or eating. I never thought of trying to apply that kind of motivation to an organization. If only there were some way…
“A Theory of Fun in Game Design” by Raph Koster touches on this: the rush from mastering a pattern of play (or a pattern of reality) is an opiate reaction (endorphin, meaning endogenous morphine). Joy-seekers are natural hop-heads, risking straightforward rewards for the rush of discovery.
Which also explains why opiates — opium, morphine, and heroin — are so attractive to and destructive for smart people. It’s literally joy in a needle, without the normal prerequisite that you learn something.
Hmm… I need to spend some time studying babies.
“Curiosity killed the cat!”
That’s why they were given nine lives. And even with eight lives gone, the cat will still be curious.
“Getting ahead humor is usually in-group humor at the expense of out groups. It too is often cruel.”
should have been “Getting along…”
I liked the three-way split “pleasure/happiness/joy” and the exploration of ‘if corporations are people, what kind of personality do they have’.
Harari (link below) sees the scientific revolution and the ensuing academia/military/industrial complex as precisely taking advantage of unexpected and disruptive advances in technology by funding curiousity. Darwin may have been very curious about the origin of species, but his expedition on the Beagle was funded by the Navy for strategic reasons.
http://www.amazon.com/Animals-into-Gods-History-Humankind/dp/1478237856/ref=sr_1_1?s=books&ie=UTF8&qid=1363265065&sr=1-1&keywords=harari
I’m willing to believe that organizations currently have no curiosity, but it raises the Searle-esque question: “If an organization had curiosity, how would we know?”
Part of the problem is that our rational selves have a tendency to state value in measurable terms. Economists turn ‘being happy’ into ‘maximizing utility’ because the latter is more amenable to analysis. That’s true of us all. Even if our actions show different, our narratives tend to emphasize the quantifiable, the tractable. ‘Curiosity’ is too ineffable for them.
The tractable is especially powerful in collectives. It’s hard to imagine people paying for something that doesn’t satisfy their particular curiosity. It’s hard to prioritize our respective curiosities. So we emphasize more quantifiable narratives instead.
Do you think your red blood cells care that they’re feeding your curiosity? The brain’s probably justifying their efforts in the name of risk avoidance or some short-term gain.
I don’t think the question is that deep.
The observed behavior would be unmotivated seeking of certain kinds of knowledge. You then eliminate other motivations one by one (like bragging rights or vanity…) and if all are eliminated, you’ve got curiosity.
For example, I think IBM doing Deep Blue and Watson is a case of curiosity. The business case (supercomputing metrics, PR, marketing, bragging rights) don’t seem enough. The only reason I am reluctant to attribute the curiosity to IBM itself rather than to the execs who sponsored it/researchers who pulled it off, is that it seems one-off rather than a systematic behavior.
The Big Data design principle of “collect everything even if you don’t know what you’re going to do with it” seems to me like a partial implementation of systemic curiosity.
Ok, let’s accept that definition of curiosity. There’s still the necker cube of viewing something as a collective vs organism. Collectives must justify curiosity to their minions.
Conjecture: Collectives of sentient agents can’t themselves become sentient, because the conflicting interests of their constituents prevent curiosity.
I don’t think sentience implies agency. Conversely, I don’t think agency requires sentience.
In fact, I think non-sentient collectives can generate internal dynamics that sort of short-circuit the agency of the parts. I think the Web is doing that to us right now. Mindless-click-addiction etc.
Sorry, didn’t mean to bring up agency. Let me rephrase my conjecture: “Collectives of self-aware beings can’t themselves become self-aware, because the conflicting interests of their constituents prevent curiosity.” Implicitly assumes that self-awareness requires curiosity.
I prefer ‘self-aware’ because it can be lost by things like mindless click addiction.
I think the gods can still feel joy due to their ability of metamorphosis, whereas our own narratives revolve around personal identity, progression and maturation. While variability is still closely related to a single continuous identity, whose properties are changing, metamorphosis is the unity of two or more separate identities within a temporal progression.
Curiosity can become attached to this or even obsessed with the horizon of ones own identity and instead of being a source of joy it becomes a cause of grief. The grief of not feeling a reconfiguration of ones own world or inner self but only the shallow variability of the same old identity is already the experience of the joyless immortals. So I wonder if curiosity needs some sort of spiritual guidance, not in the sense of positive teachings but one of dietary?
“I think the gods can still feel joy due to their ability of…
Oh! Yeah? You’re privy to the feelings of “the gods”?
Could you tell us more?
If you’re going to dredge up and co-opt an old meme, I humbly suggest changing the wording to make it more reflective of the original: “All your fates are belong to us.”
I loved the many original quotables sprinkled throughout.
I am curious why there was an elaborate refutation of curiosity being about pain avoidance or efficiency–are there strong arguments stating so?
Can’t recall where but I have heard one interpretation of Lila as curious exploration of possibilities by the One Consciousness–in that framework, curiosity would rate way above survival benefits :-)
Never thought of that, but yes, that reading of Lila makes complete sense.
Venkat,
I have been reading your posts for sometime now but never commented because most of the time I couldn’t comprehend all that you said or because I didn’t feel competent to comment.
But this post resonated with me because I have been thinking about what it means to be curious for an organisation. I work for a big organisation which is in terminal decline but was considered ‘innovative’ for a long time. I completely agree with your suggestion that “there is no such thing as a curiosity-driven joy-seeking corporation”, but I would contend that it is because fundamental drive for an organization is survival (much more than for a modern individual where survival is no more a problem atleast in the pure animal sense). The concern about survival doesn’t gel well with curiosity because when you are curious, you are not necessarily worried about survival.
Such survival worries are even bigger for listed organisations and hence they indulge in ‘pocket’ innovation but the moment market forces start the squeeze, these pockets are first to be killed.
My own thinking on this is that one way for an organisation to be curious is to consider recurrent death as a very viable alternative. In some cases, I would actually go one step further and say that actively seeking ‘death’ (as apposed to survival) may be an even better way to make an organisation curious.
Cheers
Yes, “recurrent death” is a good starting point for exploring curiosity in corporations. That’s been my own starting point too, mortality etc.
And thanks for commenting after reading for a long time. You really shouldn’t let lack of comprehension or competence stop you. I don’t comprehend half of what I write myself, and if I stuck to areas where I felt competent, I’d probably have nothing to write about.