This is the fourth of a series of posts on football and decision-making.
We’ve seen how the action on the field depends on a categorize-and-act process called procedural decision-making, and how the other team is trying to make the categorization process hard to do.
This week, I’d like to look at how players learn those categories, how players learn what actions to take in what situations.
That leads us to the playbook. An NFL playbook is a huge document of thousands of strategic plans, each of which has multiple contingencies. The complexity of a professional NFL playbook is daunting. Memorizing the playbook is even harder than it seems at first glance because the description of the playbook is semantic, declarative, and depends on neural systems that drive deliberation, but, as we said in the first installment of this series, action on the field is procedural. A player on the field isn’t seeing a bunch of circles and squares with lines marked of where to go.
These systems depend on different brain structures, and the information is stored in those brain structures in such a way that they are accessed differently. (Declarative and deliberative representations are stored so that they can be used to flexibly simulate what might happen, so that they can be used to “run the world forward”. Procedural representations are stored so that they can be used to categorize the world and identify the right action at the right time.) So there needs to be a way to translate from one system to another.
That translation depends on mental and physical practice.
Imagination engages the same structures that perception does. When you imagine a face, the neurons in your secondary visual cortex that represent that face become active. When you imagine a song, the neurons in your secondary auditory cortex that represent that song become active. And when you imagine a motor action, the neurons in your motor cortex become active. This means that imagining the play, thinking about what it would look like from your perspective on the field, engages those same neurons that you’ll need when actually on the field.
When primates (whether they be monkeys or humans) watch another primate (monkeys, humans, players on another team) taking an action, the neurons in their motor cortex that encode taking that action become active. In neuroscience, these neurons are called “mirror neurons” because they were discovered in monkeys watching people taking actions, and were recognized as “mirroring” the actions. An interesting question is whether this “mirror” property of these motor cortex neurons is something special that happens when watching other primates, or whether this is just the motor cortex imagining how it would take that action. Nevertheless, watching a play executed on tape almost certainly engages the perceptual and motor neurons needed during the game, which would help in this translation from the text of the playbook to the action on the field.
As noted in the first installment of this series, the Procedural system learns slowly, through a combination of mental and physical practice, and through still incompletely understood processes that occur during sleep. It’s not clear yet what the brain is doing during sleep, but we know that procedural behaviors become more stable and more flexible after sleep. We also know that during these sleep periods, the brain is playing through these memories, replaying the sequences and procedures. Current models have suggested that the brain is simulating variations on the procedural behavior in question to find ways to store it so that it can be executed quickly, reliably, and with flexibility. Because it takes time, this process takes days or weeks to settle in. Long term practice over the course of years almost certainly provides expertise, which enables a player to accurately respond with very complex responses to specific situations.
A lot of sports are about the kind of motoric precision that only occurs with both tremendous practice and awareness of the moment. One of my favorite moments in football is a spectacular catch on the sidelines, where the receiver stands on his tiptoes so that his feet stay in bounds at the moment of the catch. This kind of a catch is only possible by a player who has the practiced motor precision to recognize the flight of the ball (to catch it) and the practiced motor precision to recognize the position of the sidelines (to be sure not to step out of bounds).
 This is the key to the concept of brain-machine interfaces, in which neural detection technologies measure imagined motor actions. These neural detection technologies can be electrodes actually in the brain (as used, for example, by the BrainGate system) or EEG systems reading changes in oscillations that occur as motor cortex performs or imagines actions (as used, for example, by MindFlex or Force Trainer toys).
 Whether these replays are what we know of as “dreams” is still unknown – replays have been best studied in non-human animals (who can’t tell us whether they are dreaming), while dreams have, of course, been best studied in human animals (who can tell us what they just dreamed). Evidence is mounting, however, that dreams really do depend on playing through the cortical representations of what is being dreamed. Just as imagining a face engages the part of the cortex that represents faces, dreaming about a face also engages the part of the cortex that represents faces.
 Because of the way memories are stored across many neurons, memories can interfere with each other – storing a new memory can change an old one. Current models suggest that one thing the brain is doing while consolidating memories during sleep is finding ways to store the new memory without undoing the already stored memories.