Saturday, December 16, 2017

Buddhism and Evolutionary Psychology

Robert Wright’s latest book brings together two big ideas: 1) Evolutionary psychology and 2) Theravada Buddhism.  It’s pretty wonderful as I’m very familiar with both and yet the connection hadn’t occurred to me.

When I read Wright’s The Moral Animal a decade ago I was reassured that people can both accept evolution in its full force and not worry about falling wholesale into immorality.  There is plenty of evidence for cooperation and reciprocal altruism in nature – this is a theme I’ve explored in this blog years ago and one that he extended with his second book NonZero.  Now he’s written Why Buddhism Is True binding evolution, psychology, and the Thai-Forest Buddhist philosophy/practice. Buddhism, he summarizes “has been studying how the human mind is programmed to react to its environment… Now, with Darwin’s theory we understood what had done the programming.” (p 224)

There are many kinds of Buddhism (“Zen is for poets, Tibetan is for artists, and Vipassana is for psychologists” ) and some carry supernatural beliefs like reincarnation which Wright and I reject.

Vipassana meditation, from Theravada Buddhism, is sometimes called “mindfulness” meditation and has a great deal of support in the books I read.  I meditated regularly for most of a decade and intend to return; I’ve  never doubted  the value of sitting quietly and watching the mind.

First, a little background, the basics of evolutionary psychology.  Our brains were groomed over 3.5b years to compel us toward things that increase the probability of passing on genes and steer us away from threats.  Of course that was a very different environment than we have now.  Things that worked well enough were hardwired and things that didn’t were combed away.  There was no premium on accuracy so we are beset with delusions, illusions, and impulses, all nicely patched up with convenient rationalizations. 

This hardwiring becomes a problem when the environment changes faster than the genes do.  Simple example: once storing enough calories was a pretty big problem so craving fat and sugar was selected for.  When fat was scarce, that helped.  We still have the craving but now we put on dangerous weight.  Offense triggers aggression?  That worked for us once, now we get stupid road rage.  Impulses are not always the best guide.

And while we get pleasure from satisfying our impulses it doesn’t last very long.  Why should it?  There’s no traction in satisfaction -- we have to reset quickly to be ready for the next impulse.  Studies have shown that even after winning the lottery people soon go back to feeling a “normal” amount of happiness again.

Thoughts can inform emotions, that’s what cognitive behavioral psychology has shown us.  But emotions, Wright says, are the real drivers of action and the prefrontal cortex, our "thinking" and most recent part of the brain, then rationalizes the action.  This is not intuitiv. We feel as if “we” simply make decisions and act on them.  Brain scan studies show the brain signals movement before the person herself know she's going to move her arm.  On seeing a stranger the amygdala announces within 0.2 seconds “attractive=friend” or “stranger=danger” before our prefrontal cortex -- and our awareness -- registers “thing.”  In a world of snakes, cougars, and warring tribes this reactive module was useful so evolution locked it in.

Wright then pointed out the next logical thing, and I was surprised that it hadn’t occurred to me:  Emotions can be true or false. This goes against popular psychology but if the feeling of desire (groomed by evolution) says “eating this bag of chips is good for me” -- and it’s really not -- your desire is lying.

Now a little background in Buddhism, there are three truths to existence  1. Anicca, 2. Dukka, and 3. Anatta.  Nothing is permanent, there is suffering, and there is no self.  The last one probably requires a little explaining.

What are the fundamental qualities of self?  Buddha started with the proposition that to be “self” it must have some consistency through  time and it must be something that we control.  Both of these concepts are implicit in the sentence “I did that.”

The Buddha said there are five aspects to our existence.  1.  the physical body and its organs, 2. our basic feelings, 3. perceptions of things, likes vison and hearing, 4. mental formations like thoughts, habits, etc, and 5. consciousness which is the awareness of the other four.  That’s a real simple summary but the question becomes which of these is something that is persistent through time and under our control -- the answer of course is none of them.   Meditation helps you see that one by one.   

Hence, no self.

So why do we have a “sense of self?”  It’s good for the organism to look after its body, the discrete package that carries its genes.  In fact the body is probably the most important thing, you want to wrap it up with a protective sense of self.  If you don’t feel that there is something real special about what’s inside your skin you won’t have a reason to protect it.  Good thing, too, if we’re going to live, right?

Of course that's impossible for a human to easily accept, but to get a taste of no-self try this.  Meditate and focus on something that you'd normally consider part of you -- a thought, a desire, a pain ...  When it becomes the object of your attention it seems to become something other than you.

The brain encourages use to recognize a special "self" which we fiercely protect.  That makes evolutionary sense. Wright wonders whether self should be so discreetly bounded.  In meditation he noticed that a pain in his foot sent a signal to his brain – “inside = me.”  A bird chirping sends a signal to his brain – “outside = not me.”  But, he wondered, in what sense is the cramp different from the song?  They both are signals, both are processed and interpreted by the brain.  Both can affect our outlook.

The Modular Theory of the Mind is pretty compelling, I thought.  It goes something like this.  The mind contains competing modules, each of which stimulates the organism toward taking some sort of action.  The strongest module wins.  Plenty of studies show how we can easily fabricate reasons for doing something when we actually had no reason whatsoever.   

A quote shows a good example of how this can work: “A long-term module may generate a sense of guilt when you reach for that chocolate bar. It may also give you a feeling of pride when you resist the allure of chocolate.  On the other side of the contest is the chocolate lust generated by the short-term module.  But the short-term module may have subtler tactics as well.  Is it, perhaps, the module that dredged up the memory of that article about the long-term benefits of antioxidants?  It just thought the long-term module might find that article interesting?” (p 129)

This has a practical implication. Instead of trying to overcome a habit – say, smoking – by strengthening self-discipline, maybe try instead to weaken the module that has taken control.  How do you do that?  Simply by looking at it carefully, and that’s where mindful meditation comes in.  Looking closely at the desire can turn it into an object of interest rather than a transparent impulse.  Mindful observation, he says, keeps the module from getting its reward. 

Wright uses the example of a rat which pushes a lever to get a reward.  Keep the rat away from the lever and the rat will still associate the lever with reward; when you look away he'll push it again.  On the other hand if you disconnect the lever from the reward, the rat will lose interest in the lever.  In the mind, you need only to bring the lever into your awareness  to disassociate the two.  From my experience, this can work.

Wright also explores the “essence” of things.  These are the adjectives we assign to objects -- that house is a “modest” house, or that person is a “nice” person.  Everything meaningful in our awareness is an association we have assigned to it, he says, and again, this has evolved for practical reasons.   Once you categorize something you don’t have to continually reassess it.  Attaching essence to things is something we do all day.  It's a shortcut we aren't even aware of.  

We also have an “essence preservation mechanism.” If a friend does a bad thing we consider it unusual, just as when a “bad” person does a good thing.  Still good.  Still bad.  Wright suggests  that not seeing “essence” consists mainly of not feeling intensely toward things. “Dampening of feelings leads to clarity of vision,” he says (p 165)   “… not making judgments’ ultimately means not letting your feelings make judgments for you.” (223) 

The way to see this for yourself is through meditation, he says.  It’s the practice of “fighting your creator: natural selection.”

Saturday, April 29, 2017

Algorithms To Live By (Book Review)

When I’ve recommended a book to more than a dozen people and bought a few copies as gifts, I like to distill my notes one more iteration.  So I’ll jot down here what I liked most about Algorithms to Live By by Brian Christian and Tom Griffiths.  Griffiths is a psychologist and cognitive
scientist at Berkeley and Christian is a science writer.  The book rolls out a series of optimization algorithms discovered by mathematicians and computer scientists but presented at the human scale.  These solutions help when your system crashed and they also help on your bookshelves, in your communication with friends, when parking, and in your refrigerator. 

Marriage Problem – Look and Leap
This is called the Secretary Problem but I think marriage is more interesting.  If you’re looking for “the one,” exactly how long should you play the field?  Assume you can’t go back and propose to an old girlfriend.  First decide what age you’d like to marry, count the years to then and shop around for 37% of that time.   During that period note the “best” candidate but don’t marry her but after that 37% time period marry anyone you come across who's better than that.  This will maximize your chances.  Of course it’s a bit more complicated than that…   But let’s say you could determine “best” in a reasonable way, what is your chance of marring the “best” of them all this way?  37%.

Oh so now you say you can go back to an earlier date and propose?  And there’s a 50% chance she’ll accept?  Then look for 61% of your time then leap, simple as that.  Oh, marrying just for the money, and you can measure your date’s net worth easily?  Then set a threshold at 95th percentile, marry the first who’s worth more.  But that threshold will fall as you exhaust your pool, the tables are in the book.   
By the way, best chance of getting the wealthiest this way: 58%.  See how fun it is?

Setting a Home Price
Sell a house like this, if you know the high and low expected bids and can calculate the cost of waiting.  You can calculate your threshold price, which you apply immediately  and never change.  If the range is $400k to $500k, it’s a slow market and waiting costs $10k an offer, hold out for $455,279.  The graph is in the book.

Parking
Parking a car and know the occupancy rate?  With a 99% of spaces taken, you should start looking ¼ mile from your destination.  If it’s 85% full you can drive within a half block.

Getting Caught
Oh you’re a burglar and if you get caught you lose everything?  You want to know how many burglaries to do?  Just take the chance you get away and divide it by the chance you’ll get caught, burglarize that many times, then quit.

Explore vs Exploit
How long should you shop around for new friends or restaurants (explore) and when should you stick with your favorites (exploit).  That depends on your time frame.  Finding a new great restaurant isn’t going to be worth as much if you’re about to move out of town.  So explore early on, then exploit.  There are several models you can use:  “The Gittins index and the Upper Confidence Bound ... inflate the appeal of lesser-known options beyond what we actually expect, since pleasant surprises can pay off many times over.”  I’ll leave the details for the authors to explain.

Adaptive Trials
Adaptive trials are interesting, they allow you to gradually phase out the less promising of two experimental clinical techniques.  Imagine starting 50% a-50% b until the A’s appear to do better, move to 55-45 ... and so on, there's a precise algorithm. You can drastically reduce the number exposed to an inferior treatment. But clinical trials are usually not done this way.

Sort and Search
The discussion of Sort-Search tradeoffs was great; it introduced Big-O notation.  Big-O is how processing costs change with size (n).  The example was a dinner party.  You have to clean your house once regardless of n, so cleaning is “Big-O of 1”.  Passing a dish around the table increases linearly with every additional guest: “Big-O of n”.  Each guest arriving hugs all the guests already there … that’s quadratic time: “Big-O of n(squared).”  Exponential time, “(Big-O)(2 raised to n)” would happen if each guest doubles your work.  “Big-O (n!)” – factorial time is so much worse.  That’s like randomly  shuffling a deck of cards until they happen to fall in order. 

So how does Big-O help with sorting?

Bubble sort is Big-O of n (squared).  Look at every adjacent pair of books in a bookshelf and switch them if they’re out of order, then shift over and do it again. Who would do this?  A computer would, or a slow person with bad eyesight.  In practice Insertion Sort, in which you remove all the books and place each on the shelf correctly, is not much better than bubble sort although prior knowledge might save a lot of time (M is in the middle, start looking there….).

Mergesort is the punchline.  Sort smaller batches then shuffle-sort those into bigger batches, repeat and wala.  But if you are not expecting to search much, why  bother, just use Bucket Sort: put them into categories and quit.  Save a lot on the sort,  pay a little on the search.  “Sorting something that you will never search is a complete waste; searching something you never sorted is merely inefficient.” (72)


Ranking Athletes
There’s a whole section on competition scoring, like for athletes or sports teams.  The problem with Single Elimination tournaments, where one loss knocks you out, is that while it finds the best team, “second place is a lie." Round-Robin, where each team plays every other team – Big-O of n(squared) … so many games required!  And so many boring ones. Ladder Tournaments, where each player can challenge the next best is a bubble sort.  The most popular, Bracket Tournament, divides the field in half at each stage.  It’s merge sort.  March Madness takes 64 teams to 32, then 16, then 8, then the “final four” before the determining match.  It’s Big-O(n log n).  With 64 teams to start it reduces the number of games needed from 2,016 with Round Robin to 192 games to find the best team.  But it doesn’t find the second best.

Pecking Order
Chickens have a pecking order, for real.  They use displacement  where one member just knows it’s not worth trying to compete with anyone except that one directly above or below.  So place is relatively easy to establish. If crook-beak beats bug-eyes who just outpecked you, well you just don’t have to fight crook-beak – you know you’ll lose.  That’s ordinal ranking.  In comes a newcomer.  He’ll have a rough go at first, finding his place, but then it’ll be tensely peaceful again.

Race vs Fight
If it’s a race rather than a fight, it’s not ordinal that matters, it’s cardinal scores and all so much easier.  The skier places precisely by racing the clock in a couple of runs, where the cage fighter has to take on one nasty opponent after the other.  The authors explain, “Much as we bemoan the daily rat race, the fact that it’s a race  rather than a fight  is a key part of what sets us apart from the monkeys, the chickens – and for that matter, the rats.” P 83

Caching 
There are many ways to purge memory, the crudest maybe is Random Eviction.  Maybe someone with dementia suffers this. Another method is First-in-First-Out, the oldest things must go.  Clairvoyance – using future information -- is best if you can get it, and there’s an formula for that too: Belady’s Algorithm.

When it all shakes out there are times to use each of these but generally speaking the last thing we can expect to need is the thing we used the longest time ago.  That’s Last Recently Used.  So if you always put your books back on the left side of the shelf, if you return your file folders to the front of the drawer, if you hang your used shirts to one side of the rack … that’s not a bad idea.  So those papers on top of your paper piles are probably the ones you will want to grab.  Sweet!  It's that filling drawer on its side.  With this line of reasoning, throwing your clothes on the floor actually makes some sense.  I'll tell my son. If you think about it, the brain works this way, pretty much.  Those things in the more distant past, what we haven't thought of for a long time are the ones we are likely to forget -- and are the ones we can most afford to lose.  Thank you natural selection.

Sequence
Then there is the order in which we should actually do things we need to do.  Getting Things Done, the organizational system which I try to use, recommends doing the quick things immediately.  Others will tell you to do the hardest things first, or the fun things first, or the oldest or most recent things first.

Gantt charts help optimize order of operation.  For example, when you have many loads of laundry that need washing and drying find the one with the shortest cycle. If it’s the washer do that load first, if it’s the drier do it last, repeat for all loads.  You maximize the time both washer and drier are running, and minimize your time at the laundromat.

If you have a lot of tasks with deadlines you can follow an algorithm for “minimizing maximum lateness” by prioritizing those with the Earliest Due Date.  First things first.  But if you want to minimize the sum of lateness, use Shortest Processing Time (always do the quickest task first).  If you can weight each task by importance just multiply that weight by the time required.  “Only prioritize a task that takes twice as long if it’s twice as important.” (111)

Some tasks can be given an “allow priority,” that is, they aren't high priority in planning but kick in over another when needed. We do this intuitively with bathroom breaks; when you need one that task trumps most others.  If you don’t plan this allowance into a project design it could spell trouble as the authors point out happened on Mars Pathfinder in 1997 when it thrashed just after landing.

Thrashing
Thrashing is when a system is running at full bore and accomplishing nothing.  In this case some tasks should simply be interruptible.  Switching tasks (context switching) comes with a cost.  Sometimes you can reduce the cost of context switching by clustering or coalescing tasks – the author suggests scheduling a “bill paying day” when you get out your checkbook ... once.  The GTD system helps with this by coalescing actionables into folders so you can tackle related projects together.  The U.S. mail coalesces correspondence for us.  Office hours coalesce interruptions.  Answering machines do too.  It’s interesting to look at our daily activities this way.

Bayesian Probabilities
The book goes over Bayes" Rule which combines probabilities to overcome intuitive traps explaining things.  You basically use known probabilities for hypothetical pasts, figure the chance they would deliver the known outcome, and work backward to find the most probable cause. The example given was a random pull from a bag of coins containing 9 fair coins and 1 two headed coin.  It flips heads.  How likely is it to be one of the fair coins?  Calculate the chance of a fair coin being drawn and the chance of it flipping heads that’s 90% X 50%  -- and compare it to the chance of the trick coin drawn and flipping heads (10% X 100%) That’s 45 / 10 or 4.5 times more likely to be a fair coin.  Simple when you think about it, but hard to intuit.
 
Laplace took it further.  His law predicts that if you try a lottery only once and win, an estimate of 2/3 for the portion of winning tickets is better than 100% or 50:50.  It’s always the number of wins +1 divided by the number of attempts +2. So if the bus was late 3 out of 12 times the chance of it being late today is 4/14 or 28.6%  .  This one I didn’t get, but I suppose that if 68 of the other 70 insights in the book did make sense – that would be 69/72 or 96% chance this one should too?  It’s on page 131

Longevity
Here’s a real cool simple tool.  If you want to predict how much longer something you see will last and you don’t have anything to compare it to, find out how long it’s been around and guess that.  You simply assume that the timing of your sample is random, so if it’s normally distributed the best guess for current moment is smack in the middle, top of the bell curve  That’s the Copernican Principle, good when you know nothing.  So how long will North Korea last?  Let’s see… 2017-1948 is 69.  So … 2086.  Oh my.

If your phenomenon has a known distribution you can do better with the Multiplicative Rule .  The example given was movie gross receipts.  Most movies make little or nothing and some are blockbusters.  This is a power law distribution.  There is plenty of past data and the calculated multiplier that fits this particular distribution is 1.4 so if you hear a movie made $10,000 so far … best guess is a total of $14,000.  A movie grossed $6m?  Probably it will quit at 8.4.

Here’s the difference between normal distribution which call for the Average Rule, and power law distributionss that use Multiplicative Rule: “Something normally distributed that’s gone on seemingly too long is bound to end shortly; but the longer something in a power-law distribution has gone on the longer  you  can expect it to keep going.” (140)  The third method is the Additive Rule, good for things that are “memoryless,” that is, you have priors but they don’t follow a regular pattern; these are things with a wing-shaped Erlang distribution.  Then predict a constant amount of time.  Like for a slot machine, when is it going to pay off next?  After every pull, win or lose, the prediction is n more pulls.

Poor Priors
The authors noted that humans do pretty well using these methods, but we inform our own priors (normal, power law, Erlang) by experiencing life ... and in modern times our information is heavily skewed.  We’re a more likely to hear about the guy who was killed than about those who had a normal day.  If we use our priors without understanding our information bias we’re likely to make stupid predictions.  It’s why people are afraid of flying but not driving on the freeway: a plane crash is big news.

Remember the marshmallow challenge, where kids with the willpower to wait 10 minutes doubled up, and lo, those showing such willpower did better in life?  The book offers an interesting twist. When the experiment was preceded by an adult promising cool art supplies to the kids in the waiting room -- then some kids got the supplies and some a delay and lame excuse. The stymied kids were more likely to eat the marshmallow, giving up after a short effort. It seems their priors suggested these adults were not to be trusted … might as well eat this one sweet while you can.  The authors speculate that the differences between the successful kids and less successful ones in the classic study may not be just a lack of willpower, but “it could be a result of believing that adults are not dependable: that they can’t be trusted to keep their word.”  Bad priors.

Overfitting
The discussion on overfitting was another lesson on how too much weight on prior data can go wrong.  The tendency of course is to pile on the independent variables, expecting a more predictive model if you collect more background information.  But imbedded in that assumption is that each of those things is actually a good predictor of what you’re really trying to measure!  For example, taste is overfitted by evolution, to crave fat and sugar with no end.  That used to help us but today … another bad prior.

Occam’s Razor suggests the simplest hypothesis is probably the best.  If you want to simplify your model, try adding a “complexity penalty” to knock out some superfluous factors.  Nature does it: “The burden of metabolism … acts as a brake on the complexity of organisms, introducing a caloric penalty for overly elaborate machinery.” (161)  And in a similar way the slow pace of evolution prevents organisms from overfitting their environment – that’s makes them more resilient. Thanks again.

Optimization
There is a large section on optimization.  This covers the problem of local maximums, where the only way to improve in the long run requires a period of getting worse. A step down is required before larger steps up.  You can get around this with randomized “jitters,” progressively removing random influence: “simulated annealing,” and relaxation of constraints at least early in the process.  Lagrangian Relaxation is clever – it simply moves constraints over to the cost structure. You can’t steal cars to go to work!  Lagrange would respond: “actually I can, but let's consider the costs.”

Handshakes
In a section on communication the authors point out how, by signaling receipt, the receiver influences the message.  This might be by nods or facial expressions, or “oh,” “ha,” “hmm..” – in other words: “message received!”  Bad listeners actually ruin the message because the sender doesn’t know how to proceed -- what was heard, what was not, should she speed up, slow down, wrap it up, repeat something?  This sorry of feedback is what computers do, every packed or chunk is sent, acknowledged, and the acknowledgement is acknowledged. Constant handshakes, and that way a missed message is simply  resent.

Exponential Backoff
But what if there is a breakdown in communication – what can be done about it.  The answer is brilliant: Exponential Backoff.  Each time there is a failure double the delay sending it again.  The example was in dating or friendship: Oh, so she didn’t return your text?  Wait a day, send another.  No response?  Wait 2, then 4 then 8 … Very soon you’re almost out of touch, gracefully, but not completely. This is apparently how failed password attempts work, to increase security.  It’s like a squirrel, approaching you for that peanut.  Additive Increase, Multipilicative Decrease.

Tail Drop
Here’s a nice one: Tail Drop.  When we just had house phones people calling when we were out had to call back.  They couldn’t leave a message. The early answering machines took messages (buffers) but with limited capacity.  I remember "Mailbox Full" But there's no Tail Drop with email, it's an infinite buffer. Everybody expects an answer.  Texts keep coming and the buffer never fills.  “We used to reject . Now we defer.” (226)  What have we done?

Auctions
The authors write about auctions “Sealed-Bid First-Price,” “Dutch Auction (descending),” English Auction (ascending)” and the brilliant “Vickrey Auction” which is a sealed bid but the winner plays the second highest price, not the highest.  Unlike the others, when you crunch the numbers, “in the Vickrey auction, honest is literally the best policy.”  Ebay uses Vickrey, I’ve always thought it was brilliant.



So, I’ve combed through the book for many of my own highlights for my own purposes, but I recommend you buy a copy, there's so much more. It’s superbly written, packed with insight, and easy relate to.  

Sunday, April 16, 2017

On Stoicism: A Guide to the Good Life

Those who know me might understand why so much time has passed since my last blog, it’s been a tough year!  Through it one pleasure (of many) I could count on is my bicycle commute which virtually guaranteed an hour of audio-book reading five days a week.  I'd seen Stoicism mentioned in Ryan Holiday’s Ego is the Enemy and The Obstacle is the Way so I searched for a book on this philosophy.  

Good books on Stoicism are hard to find!  Finally I stumbled across A Guide to the Good Life by William B. Irvine.  I liked it so much that I listen to it twice through and then
Buy on Amazon
bought a copy to mark up.  This entry is mainly a summary of that book, of that philosophy, as explained by the author.

Stoicism fell from the public eye when philosophers began emphasizing theory at the expense of lifestyle wisdom -- Socrates had taught both.  Modern psychology’s emphasis on exploring all emotions runs counter to some of the Stoic methods, and the term “stoic” itself has taken on a meaning that is less than attractive.  Besides, people today tend to think they don’t need a life goal (and by default become what Irvine calls “enlightened hedonists:” basically, sophisticated pleasure-seekers).
   
The core idea of Stoicism is summarized by Irvine this way “Stoicism … is a cure for a disease. The disease in question is the anxiety, grief, fear and various other emotions that plague humans and prevent them from experiencing a joyful existence." (p 238)  Thankfully, unlike the original Stoic philosophers Irvine doesn’t refer to Zeus to explain how these emotions were embedded in us.  He looks no farther than evolution which, through natural selection, built emotions into our psyche for the purpose of survival and reproduction.  Evolutionary psychology is fascinating.  But the world has changed and the impulses that once helped don’t serve us well anymore.  Fortunately, one of the tools evolution has given us is rationality and we can hijack that  to reset our own life goal.  Our life goal doesn't have to be reproduction.  Stoics would recommend tranquility.

Tranquility is not the only goal possible, to be sure.   Hedonists seek to maximize pleasure and this is probably the default goal of most everyone. We're pleasure seekers because in geologic time the things that gave us pleasure steered us in ways that increased survival and procreation.  “Enlightened hedonism” today uses rationality to weigh long and short term pleasures and costs but in the end, he who has had the most pleasure wins.  The Cynics’ goal was virtue, harmony with nature, and to reach this they advocated rejection of all conventional desire.  Stoicism is in between: not ascetic, not hedonistic.   The stoic's goal is virtue and tranquility.  Tranquility is a goal of Buddhism and Epicurianism too but they go about it in a different way. 

Unlike Cynics who strive to eliminate desire the Stoics try to minimizing negative emotions and to live a 'virtuous'  life. Virtue and tranquility go hand in hand -- both rely on rational thought.  In this book Irvine focuses on tranquility and explains how Stoics learned to increase it. Maybe he'll write another on virtue. 

The Roman Stoics Irvine highlights are Seneca, Rufus, Epictetus, and Marcus Aurelius who was the  Roman Emperor from 161 to 180.  Each of these contributed something to the philosophy and most of them were exiled. Some were killed for being philosophers.  According to Irvine Descartes was a Stoic, so was Henry David Thoreau, but for the most part the philosophy waned as philosophers turned to theory, the public felt no need for a philosophy of life, psychologists emphasized methods that ran counter to Stoic ones, and religions pressed competing ideas too.   But here are some highlights of the book.

NEGATIVE VISUALIZATION
This is one of the core techniques.  We are set up psychologically to swing from desire to desire, satisfying one, forming the next, and so on.   Each time we attain a desired object we are rewarded but the psychological surge soon flattens and we pursue the next desire. For example, Irvine observed,  people often go to the mall not because they need something but in hopes to stimulate a desire which they can satisfy... for that dopamine boost.  We are insatiable; the Stoics suggest that instead of trying to satisfy desires endlessly we would do better to manage them. And confounding things more is what behavioral economists call hedonic adaptation: We place more value on things once we own them.  If you find $500 that’s nice, but if you lose $500 that’s very bad.  To use a recent example, if you think you won the Academy Award for Best Picture 2016 and then learn it was just a mistake you’re absolutely worse for the experience.

We are driven to acquire, we take our acquisitions for granted, and when we lose them it is an outsized blow.

So what do Stoics do about this?  One important exercise is negative visualization, that is, periodically imagining losing something valuable before it happens.   It’s not actually worrying, it’s contemplation.  Doing this can inoculate ourselves so when the pet (or child!)  does die or the teapot breaks … well we sort of knew it might happen ..  and that hurts less.  We’ve reduced the negative impact of grief.  If you consider for just  a moment that your daughter might die that makes actually being with her so much sweeter and every moment more meaningful. If she were to actually die and you had appreciated her life fully you may suffer less grief - - and less regret too.  The Stoic way provides more joy while she’s alive and - -  if it comes to it - - less pain when she's gone.  All from negative visualization

Stoics also recommend what Irvine calls voluntary discomfort: sometimes taking on a loss or suffering purposefully -- not in a masochistic way but an instructive one.  Reading this I was reminded of when I told my dentist before what was basically a root canal that I would like to do it without anesthetic.  I wanted to experience extreme pain without actually suffering harm.  Doing this has several benefits: it inoculates one against the full shock when actual hardship occurs, it widens one’s zone of comfort, it helps one appreciate what we do have, and we gain confidence going forward that we can bear what life might deliver.  The flipside, also recommended, is, mindfully abstaining from pleasurable things now and then. 

There are other tools related to negative visualization...  like imagining that the bad things that happen to other people have happened to you, or trying to see bad things that happen in a more cosmic perspective. In the universe, how bad is it really that I just suffered, <fill in the blank>? 

FATE
The serenity prayer suggests we accept the things we can’t change, change the things we can, and have the wisdom to know the difference.  I’ve heard that all my life, but it was a bit eye opening to look just a little deeper.   Nothing in the past can be changed, it’s over. Then what is the use of all regret, it’s just a useless negative emotion!  A lot of anger goes out the window too. We can still learn from the past, but that certainly takes a lot of sting out of it.

More strangely, though, is that we can’t actually change the present either – the one we live in.  As soon we take an action that present has past.  Therefore, if we follow this line of reasoning, there is no anxiety or unhappiness about the way things are, in an immediate sense.  That’s very calming; we really can't entertain thoughts on how things right now could be better.

And then there are those things over which we have some but not complete control -- Irvine's simple example is a tennis match.  As much as you’ve practiced, as well as you play (both are things you can control) … you don’t control the skill of your opponent and therefore you can’t control the outcome of the game.  If you set a goal of winning the game and you lose it you will be disappointed.  Even if you win you likely will have suffered anxiety along the way.  Stoics recommend not trying to win the game, but rather set an internal goal … “playing your best.”  That goal you can control completely.  You can lose the match and still feel satisfied and accomplished if you've played your best... and you have no less chance of actually winning.  Opponents become teammates. With that internal focus, which you can control, you're likely to play better and may well succeed and win.

INSULTS AND OTHERS’ OPINIONS
Other lessons relate to what we think of others and what they think of us.  Insults are interesting in this regard.  They’re not really injurious in themselves, like a punch in the face would be.  But they hurt only because of what we make of them. The advice?  It’s very clever.

First, consider the source; if it’s someone whose opinions you don’t respect, their insult can be dismissed out of hand. Irvine suggests considering it like a dog snarling at you – you might make a mental note of it but you don’t go through your day worrying about the fact that the dog doesn’t like you! But, if it’s someone you respect or someone who is well-informed about the topic at hand, look straight past the insult - - perhaps you’ve found a teacher. 

Second, if the insult is true and obvious, then it’s just a statement of fact.  I’m bald … ok, so?   But if it’s not true perhaps it should be corrected, like a parent correcting a child's bad behavior. It can be done kindly.

Third, if you choose to respond, the best way is often with humor because that can erase the intended sting.  Self-deprecating humor is best as it can send an ironic message of confidence when the insulter intended the exact opposite effect.  If the quick wit is not not at hand, don’t respond at all.  Either way you signal your disregard for the person's opinion rather than take them on.  I like that. 

GRIEF
Already we know that negative visualization can diminish grief because you’ve imagined the loss.  But one can also practice retrospective negative visualization, that is, imagine not having ever had the thing which you've lost.  In essence, “be happy with what you have had.”

If grieving the death of a loved one you also might do well to wonder how much pain the person would have wanted you to suffer.  When someone else is grieving, the Stoic advice as expressed by Irvine was don’t catch the grief deeply, but you might feign grief if it makes them feel better.  Personally, I think actually sharing the grief, even if briefly, could be a good thing -- also a form of voluntary discomfort. 

In any case, although it's a negative emotion they say the griever should grieve.  Seneca wrote to a friend “Let your tears flow but let them also cease, let deepest sighs be drawn from your breast, but let them also find an end.” (p 154)  Here the Stoic advice runs against popular grief therapy which often advises lingering much longer in grief, fully embracing it.

ANGER
Anger is not only “anti-joy” but also is also irrational.  “Reason,” Seneca advises, “will never enlist the aid of reckless unbridled impulses over which it has no authority.” (p 160)  Yet feigning anger can serve a purpose -- signaling that you are likely to act irrationally (and therefore are unpredictable). But actually abandoning reason by way of anger is not recommended by Stoics.  

So how do you reduce anger?  Resist thinking the worst of people, don’t get so comfortable that you’re easily angered, and notice how much more the anger causes your own suffering than the trigger for the anger itself did. With anger we punish ourselves.  Attempting to view the offense from a more cosmic perspective can diminish its effect.  And if you can, turn anger upside down by finding humor. Laugh.

And apologize later … that matters.

FAME AND FORTUNE
Fame should be avoided -- we basically should be indifferent to social status, they say.  Irvine paraphrased Epictetus: “If we make it our goal to please others, we will no longer be free to please ourselves.”  Just as insults and criticism don’t have to infect us with negative emotion, we should become impervious to praise and compliments too.  It’s a slippery slope, relying on others in this way; better to set those internal goals.

Interestingly, luxury is a bit different (although various Stoics apparently disagreed on how much one should enjoy it).  Basically, wealth can be a fine thing as long as you control it, and not it you.  Its dangers include 1. not being prepared to lose it and 2. the loss of one's ability to enjoy simpler things.    

It makes sense that Stoics are often wealthy.  They set internal goals that can result in external successes just the same, and they are happy with a simple lifestyle. They may well earn an excellent wage and not spend nearly as much as others.

EXILE, OLD AGE, AND DEATH
An interesting short chapter dealt with exile -- probably because most of the Stoics mentioned were exiled, sometimes repeatedly.  From the unflappable Stoic perspective exile is just a change of place.  Irvine brought this home by comparing exile to being “banished” to a nursing home, a fate many do still suffer today.  He explains, using Stoic principles, how one near the end of life might suffer fewer and less intense negative emotions and still find joy, if they know how.

---------------------------------------------------------------
The book is nicely written, it appears to be an excellent synthesis of the original Stoic literature with modern applications. It's nicely indexed and well referenced and the audio edition is well read.

Since writing this summary I've read a translation of Marcus Aurelius - - The Emperor's Handbook: A New Translation of the Meditations. I think Irvine did well - - his book was better IMO! Most of his observations. I think, were drawn from Seneca - - I haven't read Seneca yet but will.