It's a Jungle in There: How Competition and Cooperation in the Brain Shape the Mind (21 page)

Star basketball players and Olympic skiers make perceptual-motor creativity especially clear. At the other end of the quality-of-performance spectrum, young children who don’t yet know how to get dressed remind us that everyday skills take practice. People who have had strokes or other neural or muscular insults teach us the same thing, as do modern robots, not by dint of their great abilities but by virtue of their woeful inadequacies. Robots today can’t do what most young children can do routinely—climb trees, tie simple knots, or scamper through jumbles of toys strewn over bedroom floors.

Regarding robots, consider two displays of computer prowess that wowed the world. The first was IBM’s “Deep Blue” computer, which in 1997 played chess against the best player in the world, Gary Kasparov. To the chagrin of many, Kasparov lost the match. Then, in 2011, another IBM computer called “Watson” beat Ken Jennings and Brad Rutter, the two best
Jeopardy!
players in the world before Watson came along.

What Watson and Deep Blue did was impressive. These computers displayed remarkable abilities to solve problems when symbols were manipulated and verbal information had to be looked up as quickly as possible. On the other hand, few people paid attention to the fact that Deep Blue couldn’t set up the board or move the pieces. Watson, likewise, couldn’t understand the host’s speech; the test items were presented to it as digital text. Nor could Watson make its way to the studio, doing things like opening a door, walking down the street, hopping on a bus, grabbing a seat, getting off the bus, and sprinting to the studio in time to record the show. These physical tasks may have been considered too humdrum to warrant IBM’s attention. Still, if these tasks had been included in the repertory of things the computers had to do, there would have been no match between the humans and the IBM machines. The humans would have won hands down.

Recognizing that the planning of skilled physical action doesn’t just happen but instead relies on considerable intelligence (not to mention quick nerves and resilient muscles) raises the question of what processes allow those skilled actions to appear. How are movements planned and controlled? And how, relatedly, is stability managed? How, in other words, do you stand while riding a crowded, bumpy subway, and how do you stand your ground while waiting for a traffic light to turn green in a howling wind? Are the processes that allow for these impressive feats of motor control compatible with the jungle principle?

One way scientists have studied motor control has been to analyze errors. By examining mistakes, they have pursued the idea that the
faux pas
people commit betray the inner workings of their nervous systems.

Suppose, for example, that you’re groggy and intend to turn off a light but instead turn off a radio. This error tells you something about how you control your behavior. Similarly, if you’re distracted and answer someone’s question about what you’re doing by telling them, “I’m writing a mother to my letter,” that slip of the tongue reveals something about what’s going on in your mind.

Errors like these, where you say or do something that doesn’t line up with what you intend, share an important feature. They reflect competition. Consider the example of turning off the radio rather than the light. Were it not for inner “demons” campaigning for radio termination, there’d be no reason to expect you to turn off a radio when a light needs dousing. Likewise, for the mother-rather-than-letter example. If it weren’t for inner “elves” campaigning for saying “mother” when the next noun to be said was “letter,” there’d be no reason to expect you to say “mother” too soon or “letter” too late.

Mistakes of this kind have fascinated psychologists for over a century. Such bloopers can be amusing but, more important, they can be revealing. Sigmund Freud believed slips of the tongue reveal pent-up urges.
⁵
According to Freud, if a man says “mother” too soon, it may be because he has an unresolved Oedipus complex. For language scientists, a more mundane account suffices. The mental representation of “mother” happens to be assigned to the wrong slot in the cue of forthcoming words, possibly because that mental representation competed too vigorously for the slot already reserved for another noun awaiting pronunciation. The fact that “mother” changes places with “letter” suggests that competition is strongest among words of the same grammatical class. Nouns exchange with nouns, verbs exchange with verbs, and so on. This fact grants grammarians gravitas they might not have enjoyed otherwise. Teachers of grammar who implore their students to learn about nouns, verbs, adverbs, and adjectives can take heart in the fact that those grammatical classes are real mental categories, not just artificial divisions used for socializing kids about rules for writing.
⁶

Errors don’t only involve speech. They also involve nonverbal behavior. When someone accidentally turns off a radio rather than a light, the mixed-up actions share the “turn-off” feature. If you decide to turn off a light, the associated intention can trigger a different action that satisfies the same general need. Inner demons that contribute to turning off the radio jump up and down, as it were, when information comes in that something needs terminating. If their activation is too high relative to the activation of demons associated with light extinction, the radio rather than the light is turned off.
⁷

It’s tempting to say that when you screw up, you do so because a little supervisor in your head temporarily loses it. Think of an editor who checks the text that’s supposed to appear in a magazine. Reporters supply stories to the editor. The editor needs to ensure that every word is correct. When
a mistake appears in print, it’s the editor’s fault. The buck stops with him or her.

Is there an editor in your brain? From the perspective of parsimony, it would be better to say there’s not. Saying there’s an inner evaluator begs the question of how that agent knows what’s right and wrong.

But can the absence of errors, and the occasional presence of errors, actually be explained without appealing to an inner editor? A cognitive psychologist at the University of Illinois, Gary Dell, argued that they can be. He devised a neural-network model consisting of nodes that interact more or less as neurons do. Excitation passes between some nodes while inhibition passes between others.
⁸
According to Dell, the nodes that pass excitation or inhibition to other nodes are identified on the basis of which sentence is supposed to be produced. By Dell’s way of thinking, the reason someone makes the mistake of saying, “I’m writing a mother to my letter” is that the “mother” node out-excites (or out-inhibits) the “letter” node. This may happen every so often by chance alone.

The sort of neural network Dell endorsed can support nonverbal behavior as well as verbal behavior. Therefore, just as the model needs no editor to explain slips of the tongue, neither does it require an editor to explain slips of the hand. I made such an error on the morning I first wrote this. Just after waking, I turned on the radio and then turned on the water to brush my teeth. The radio volume was too low, so what did I do in my half-asleep stupor? I turned up the tap! I knew I wanted to turn something up. I wanted to make something more intense. The thing I applied turning-up to was incorrect, however.

“Now wait a minute,” you might say. “Shouldn’t there be some sort of internal supervisor who helps with things like self-control?” Holding back is often crucial. Without self-monitoring, civilization would crumble. People would tell others off more often than they do, they would hit others more often than they do, they’d rape others more often than they do, and so on. Given the clear introspective evidence all of us have that we can control ourselves, is it right to say no editor resides in our brains?

Suppose for a moment that there
is
an inner editor. You might have one if you’re a reasonably self-controlled person who checks what you say and do. Freud named the editor the
superego
.

One reason to grant credence to an internal editing process is that sometimes you know you’ve made a mistake even when you’re making it. Consider releasing a basketball at the free-throw line. You may know as you release the ball that the ball won’t go in. Experts viewing free throws can make such judgments even when they see the ball only at the moment of release, so it’s
plausible that the shooter him- or herself also knows whether a shot will go in from the moment the ball leaves his or her hands.
⁹

A more subtle phenomenon that reveals the same principle is that during typewriting, incorrect keystrokes are made less forcefully than correct keystrokes.
¹⁰
From this simple observation an interesting implication follows. If you respond less forcefully with keystrokes that are wrong than with keystrokes that are right, then at some level you know those keystrokes are wrong even while they’re happening. Some internal process tries to inhibit the keystroke, but the inhibition comes too late.
¹¹

Another source of evidence that there can be internal checking of actions comes from the fact that some verbal and action slips occur because, ironically, you check your performance
too
carefully. If you monitor how you’re behaving too much, the vigilance can interfere with how you perform. This point is well known to anyone who suffers from stage fright. Worrying about the way you perform is a kind of editing. When the self-monitoring gets too overbearing, the wrong mental creatures are turned on and may exert inhibitory effects on the ones whose outputs are actually adaptive.

Another hint that self-monitoring can be harmful comes from research showing that focusing on your own behavior can hurt your performance. Athletes and musicians often say it’s better not to think too much while performing, and recent research bears this out.
¹²
Sian Beilock of the University of Chicago and her colleagues asked experienced golfers to putt in two conditions. In one, the golfers were asked to attend to extraneous stimuli that diverted attention from the putt. In the other, the golfers were asked to direct attention to the step-by-step nature of their putting. Golfers in the step-by-step condition did worse than golfers in the attend-elsewhere condition.
¹³

How do these results bear on the question of whether an editor is needed to explain errors? They show that inner supervision can occur and can be detrimental. On the other hand, saying that inner supervision can happen doesn’t imply that there really is an intracranial being whose job is to legislate. Inner creatures can have collective effects that give the appearance of editing. None of the phenomena reviewed here actually needs an editor to explain them. Appealing to neural creatures that become more or less active suffices to explain the effects just covered.

When neuroscientists have delved into the neural dynamics of movement preparation and control, they have found that the brain relies on a process
more akin to democracy than dictatorship. Consider research from a team of neuroscientists led by Apostolos Georgopoulos, working at the time at Johns Hopkins University.
¹⁴
Georgopoulos and his colleagues made recordings of neurons in the motor cortex of monkeys. The monkeys moved one hand to each of a number of targets in a circle from a central home location. When a target appeared, the monkey was supposed to move its hand to the target in order to get a reward. All the movements were made in the horizontal plane.

By recording neural activity in the motor cortex, Georgopoulos and his colleagues demonstrated neural democracy. Their discovery was embodied in a principle called
population coding
.
¹⁵

Population coding works much as an election does (
Figure 9
). In the case of manual positioning, it relies on several facts. First, individual neurons fire most strongly to particular directions of movement. Second, all the cardinal directions (0 degrees, 45 degrees, 90 degrees, etc.) have neurons that prefer them. Third, particular directions of movement arise from the votes of all the neurons.

Consider each of these facts. Regarding the first one, that individual neurons fire most strongly to particular directions of movement, Georgopoulos and his colleagues found that individual neurons in the motor cortex fired the most when particular directions of arm movements were made. The more the direction of arm movements deviated from a neuron’s preferred direction, the less the neuron fired.

FIGURE 9.
Population coding. Each solid arrow corresponds to a neural pool. The angle of each solid arrow shows the favorite direction of a neural pool, and the length of each solid arrow shows the strength of activation of that neural pool given a target (the dot). The dashed arrow shows the sum of the solid-arrow vectors (the population-code vector), pointing in a direction that none of the neural pools prizes the most.