I’ve spent the last five months conducting research propelled by the belief that these questions can be answered, and that the answers will be rooted in solid science. Finally, a clear picture is starting to emerge. This picture binds together thinking I’ve encountered from motivation theory, behavioral economics, emotional intelligence research, neuroscience, machine learning theory, and psychoanalysis. Explaining all that I’ve learned would take a book and I haven’t finished putting the pieces together yet. However, I’m going to start trying to explain what I’ve discovered. I have a bunch of thoughts to share with you, so I’m going to break the journey up into several posts. Hopefully, it’ll prove both useful and entertaining.
I Suck and I Love to Fail!
One of the greatest lessons that improv teaches is that if you don’t trust yourself to do something perfectly, do it with glee instead. Great improv teachers like Kat Koppett talk about ‘celebrating failure’. Troupes all across the world yell out ‘I suck and I love to fail!’ to each other with huge grins on their faces before going onstage. Trainers everywhere talk about ‘mistakes as gifts’. It all amounts to the same thing: making the most out of getting things wrong.
This idea might sound cute and more useful in a comedy show than the workplace, but I’m going to show you how it can help us overcome arguably the most fundamental limitation of the human mind. I’ll show you how it can change both the way we think, and what we’re capable of achieving.
The Brains of Bugs
In order to explain, let’s cover a little background about how thinking works. And, for starters, let’s look at the most basic kind of thinking we know about: the sort of thinking that happens in simple creatures like bugs and slugs. Unlike us, these creatures don’t spend much time worrying about bank balances or planning vacations. Instead, they just react to the stimuli they sense around them. It turns out they are capable of learning, but only in the most basic way possible: by associating certain stimuli with either pleasure or pain so that they know whether to seek them out or avoid them in the future.
AI researchers have spent plenty of time studying the brains of these kinds of creatures and it turns out that the root principle of how they work is pretty simple. Each neuron in a simple brain can be thought of as a bit like a voter taking part in an election.
Let’s imagine a contest a bit like American Idol. In this contest, all the viewers are watching over the internet and the footage that each voter sees is a bit different. With each round, the voters are all shown footage of the contestants, and on the basis of what they see, they vote. The votes are tallied up and one of the contestant wins.
However, what the voters in our contest don’t know is that their choices have big implications. Lucrative music industry deals are being won or lost on the basis of who gets picked. So after each vote, the entertainment company that hosts the contest changes the footage that each voter will get to see in the next round by just a little bit, so as to make sure that voting in future will be a bit more favorable to their interests. Those who voted for the ‘right’ candidate get to see more footage. Those who voted for the wrong one only get to see some edited highlights.
In a real insect’s brain, instead of singers to choose between, we have decisions like ‘remain very still’ or ‘flee from the spider’. Instead of an entertainment company we have the consequences of those decisions, such as ‘going unnoticed’ or ‘losing a leg’.
After each pleasant experience that our bug has, the voting neurons that caused him to have that experience get reinforced--they get to see more footage. Those that tried to guide him away get weakened--they get the edited highlights. It works the same way when our bug experiences pain. Those neurons that cause him to walk into an unpleasant experience get weakened. Those what would have helped him make a different choice get reinforced. In this way, and with a little luck, our bug learns to make better decisions over time. With luck he gets to breed before turning into somebody’s snack.
While this picture is, of course, a sweeping generalization about how simple brains work, hopefully it makes it clear that pleasure and pain use similar mechanisms. In both cases, some neurons have their connections strengthened while others are weakened. So far so good, but human brains aren’t like the brains of bugs. People, for the most part, are smarter.
The Human Difference
There are lots of differences that we could talk about between the brains of bugs and those of people. It’d be easy to get caught up in a conversation about consciousness, for instance, or creativity. But the broad, distinguishing difference, I would argue, between simple brains and brains like ours is that our brains can make plans.
Somehow, we’re capable of behaviors which don’t head us directly towards pleasure or away from pain, even though, at the end of the day, we like pleasure at least as much as your average insect. We’re even able to devise goals like visiting the moon. In order to achieve such goals we have to be able to build tools, cooperate, imagine, and reason out enormous problems, all without any kind of direct sensory reward, like, say, a large quantity of chocolate cake. How does the brain do it? I propose that the planning process in our brains happens by a process we might call recursive goal matching.
Recursive goal matching means that when a person is considering some reward he’d like to get, his mind identifies interim scenarios that will help him reach that reward. It then treats those interim scenarios as goals in their own right and tries to build chains of behavior that will reach those goals. Some of those chains of behavior contain interim scenarios that become new goals, and so on.
For instance, in order to get the payoff of eating the cake we’re keeping in the fridge, first we have to go to the kitchen. Visiting the kitchen is an interim scenario that becomes a goal. There’s no intrinsic reward in getting there, but in doing so, we’re helping fulfill our yummy plan. However, visiting the kitchen requires that we leave the couch and walk there. The goal of leaving the couch requires that we turn on the table lamp beside us so we can see our way. Etc.
When we successfully complete a goal scenario we’ve pictured in our mind, we receive a small jolt of internalized satisfaction. If we fail to match such a scenario after repeated attempts, such as for instance discovering that someone has locked the kitchen door to keep us away from the cake, the mind sends us a dose of frustration. Just as in the case of the insect brain we looked at earlier, our brains manage our behavior with tiny pulses of neurotransmitters that change how our voting neurons are wired, and thus, what they get to ‘see’.
For a fascinating account of how the brain constructs behavior out of hierarchies of such plans, I recommend ‘On Intelligence’ by Jeff Hawkins. However, what no book I’ve read so far has yet pointed out, though, is that the process of learning through planned behavior is asymmetrical. In other words, while achievement and pleasure have plenty in common, frustration isn’t experienced like pain.
The Magic of Failure
Just as in the case of a pleasure response, the satisfaction we get from reaching a goal reinforces those neuron connections that enabled us to get there, while weakening those that voted for other options that would have prevented us from succeeding. However, what happens in the case of frustration? When we fail to reach a goal, we have no idea which neurons to reinforce because we lack the knowledge of what would have caused our plan to succeed. My guess is that the brain has no choice but to weaken all the connections that might have been responsible. This means that unlike pain, failure is a universally negative sensation.
Let’s go over this again to make it clear. We’ll look at each emotion and the effect in the brain that it causes.
- Pleasure: Strengthen connections that led to pleasure. Weaken connections to neurons who voted against it.
- Pain: Weaken those connections that led to pain. Strengthen connections to neurons who voted against it.
- Achievement: Strengthen the connections that led to us reaching our goal. Weaken those that voted against it.
- Failure: Weaken those connections that led us down the path to the failed goal. Wait, what do I strengthen? We don’t know which neurons voted against failure, because any part of our plan might be responsible for it not working! If we get to strengthen anything, its the neurons that voted for us to not take this goal on in the first place!
Here’s an example of this principle in action. It happened to a good friend of mine. A scientist we’ll call Amy. Amy recently had a wild idea that she thought was going to change her field. She spoke to her boss about it, but her boss shrugged and pretty much dismissed the idea as impossible. So, undeterred, Amy boldly set about exploring the idea on her own.
To her great delight, it worked. Amy was terribly excited, and convinced that her discovery would make her career. She felt on top of the world, and was rightly proud of her achievement. When she carried out a literature search, she could find no evidence that anyone had uncovered the same extraordinary result. She wrote up the paper and submitted it. Then, out of paranoia because the stakes were so high, she conducted a second literature search the moment the paper was accepted. This second search turned up a paper that had uncovered the same result three years earlier! Amy hadn’t found that paper because no-one in the literature had been referencing it. She was mortified and withdrew her paper from the journal the same day.
When I learned all this from Amy, I could tell that her self-esteem had taken a hit. She spent the evening questioning whether she was in the right field and whether she was cut out for science at all. She felt like a failure, and stupid for having made a mistake. I’m sure we can all understand how she felt. But let’s take a closer look at what happened.
For starters, Amy’s result was right. The fact that she wasn’t first to find it didn’t affect that. Furthermore, it wasn’t surprising that she didn’t know the result had been found before because nobody had referenced the work. This was simply because the result wasn’t one that anyone wanted to be true. There’s a whole other blog post in here that I’ll save for another time, but broadly speaking, even in science, people see what they want to see.
Despite the fact that Amy was now one of only two people in the world to understand a deeply important result, Amy felt like a loser. This was because the goal she’d built in her head was that of publishing a paper and getting recognition for the result. Her brain had automatically matched to the end result of the process she’d anticipated, without considering the interim benefits like ‘doing good science’. Furthermore, her instinctive response to this crisis wasn’t ‘I need to do better literature searches’, but instead simply, ‘I failed’. She went from contemplating her own genius to wondering if she needed to look for a new job, all inside of about half an hour. To my knowledge, Amy still hasn’t really pursued her line of research further. She feels kind of bad about it. Being brilliant hasn’t prevented Amy from having a brain that doesn’t know how to process failure.
I strongly suspect that if we look back over our own life experiences carefully, we’ll spot situations we’ve all been in that are very much like Amy’s. I know I have. Such episodes can be hard to pick out because the brain doesn’t like to think about failure, but they’re there. In my experience, it’s often easier to catch such experiences while they’re happening. Next time you experience a sudden surge of self-criticism, ask yourself exactly what the goal is that you’re not matching, and whether it even makes sense.
So, in a nutshell, when we succeed, we get reward for a specific result. When we don’t match a goal, we get blasted for everything we’ve done recently, even if the goal we were matching against wasn’t a very meaningful one. This effect changes the kind of plans we’re likely to come up with. We’re going to be biased toward those plans that avoid failure, because failure experiences are going to be disproportionally negative.
Sure we can all ‘learn from failure’ as self-help books encourage us to do, but those very same books have to encourage us to do it because it’s not a natural part of our thinking process. Without conscious coaching, the brain usually has no idea exactly what the lesson is that each particular failure grants us.
The fact that we treat failure differently from pain can help explain why people react irrationally to the presence of free gifts as Dan Ariely describes in Predictably Irrational. Something that’s free comes without mental attachment to costs, and therefore potential failure scenarios, which makes it automatically desirable in plan building. It also explains the principle of 'loss aversion', as described in Sway by the Brafman brothers, that causes people to sometimes go to seemingly absurd extremes to avoid failure.
Is there any evidence to support the idea that this difference in learning patterns is responsible for these effects? So far, the evidence is still thin, but it’s building. Recent research has revealed that damage to the amygdala causes loss aversion to be suppressed. The amygdala is the part of the brain that’s responsible for dealing with the consequences of fear and other similar sensations. If the phenomenon of loss aversion is bound up with the process of suppressing links between neurons, just as pain is, then this is exactly where we’d hope to find the experience centered. This result is far from conclusive, but it’s a start. However, while this result is interesting, it still doesn’t say much about improv. For that, we have to look at some of the implications of how the brain processes failure.
When the brain knows that it’s failed, but not why, it has a problem. Just like any general waging a campaign, it has no choice but to invoke ‘Plan B’. Plan B, in this case, is a suite of backup behaviors designed to resolve tricky situations. These behaviors are ones with a long track record of proven evolutionary success and broad applicability. However, they’re often significantly less sophisticated than the behaviors we build via planning. Our backup behaviors are designed to get us out of trouble fast and often come with physiological knock-on effects to accelerate our responses. These behaviors are ones we might class as ‘irrational’. Which behaviors are kicked off depends on how much stress we’re already under when our plans start to fail.
This is why people become irrational when negotiations fail or expectations aren’t met. It underpins the sort of situations outlined in Crucial Conversations by Patterson et al, and connects up tidily with the themes of emotional intelligence research pioneered by Dan Goleman.
What we know from this research is that by changing how we look at failure, we can change how we respond to it. Our subconscious mind isn’t so hot at deciding what kinds of failure are genuinely dangerous, because failure, by definition, represents a lack of data. However, with a little conscious reflection it’s often straightforward to see that we’re being intuitively navigated away from situations that just aren’t that risky. Consequently, exercises that encourage the brain to treat failure as something to be accommodated and embraced mean that our more extreme ‘Plan B’ responses get activated far less frequently. By habituating a reasoned, conscious, up-beat response to failure, we stand a far better chance of coping well when something goes wrong, and this is what makes improv exercises so powerful.
By remembering to say ‘I suck and I love to fail!’ we are directly targeting and deactivating that part of ourselves that gives us stage-fright, makes us panicky in romantic situations, or gets us into fights. The more we practice that response, the easier it gets. This is what makes improvisers look so witty and fearless on stage. They look that way because they are witty and fearless. However, this isn’t because some kind of innate talent. It’s because their brains have learned to treat being on stage as exactly what it really is: just being at the other end of a room from a bunch of people sitting down.
I suspect that at the scale of whole societies, some well-applied improv exercises might go a long way toward making the world a more peaceful, rational place. Not a bad result for a trick to take the edge off unplanned performances. Admittedly, though, I haven’t said anything yet about what shape our mental plans actually take or how we choose which ones to follow. I’ll cover that next time, and that’s where it starts to get really interesting.