The Ghost in My Brain (8 page)

BRAIN DAEMONS.
In my academic field, artificial intelligence, there are two ways to simulate human intelligence: write computer programs using
human AI
techniques that attempt to model the actual processes of human thinking, or write programs using
alien AI
in which we write whatever sort of intelligent programs we feel like, as long as they get the job done, with no attempt to duplicate the way a human achieves the same goals. The former is typical of
cognitive science
, wherein we try to replicate the functional structures present in the brain. The latter captures the spirit of a device I'll use in this book, called
cognitive daemons
, wherein we simulate, and discuss, the effect of such processes on our lives, without regard to their real, underlying neural structure. That is, even without knowing how a human actually arranges to get more than one thing done at a time, using the simple computer science construct of a cognitive daemon allows us to accurately describe our feelings and behavior in a way that is easy to understand. In this way, even without proposing any neurological or anatomical models, we can nonetheless still claim that the design constraints on models of such systems, represented by my recorded experiences, are real.
*

A
daemon
, for the purposes of our discussion, can be thought of as a “little guy that wakes up when needed to go perform some task, and reports back to you later with the results.” Then he goes back to sleep (or dies off) until needed again. In the brain, what I refer to as daemons are thought processes that run independently, in the background (think, perhaps, of subconscious processing), and do their job over the course of
seconds, minutes, days, or even weeks—working on some particular subproblem, and then interrupting conscious thought (or a different daemon) sometime in the future with their results. Sometimes daemons will spawn sub-daemons of their own.
*

We cannot always control when these daemons get started; we can seldom control turning them off. A daemon tends to run its course—either solving the problem with which it has been tasked, or naturally and gradually giving up, as other problems take precedence. Daemons may be somewhat anxiety-driven, and slightly obsessive, as in “I just cannot remember that guy's name from California—the one who took us on his boat. What was that darn guy's name?”

Daemons run independently, but they can interfere with other, unrelated cognitions in at least two important ways: First, these background processes take up brain resources, leaving fewer resources for other processes, both conscious and subconscious. For example, if you are worried about your son's health, and in the back of your mind you are continually thinking about hypothetical diagnoses for his condition, you will appear preoccupied to others. You will perform less well on the other tasks in your life; you'll misplace your gloves and forget to feed your pet fish. This makes sense: there is only so much cognitive processing power to go around.

Second, these background daemons appear to communicate with other parts of the brain via an
interrupt system
. (The use of interrupts is a style of computer programming in which one program is allowed to
interrupt
the processing of another program to communicate with it.) You can certainly see this happen in your own life: While you are in the middle of making lunch, you might pause for a minute because the cheese you are slicing reminds you of a bicycle picnic with your friend Lisa ten years ago, at which you ate cheese sandwiches while sitting beside a stream. Lisa was friends with
Gary
, and
Gary, Indiana,
was where the California guy (with the boat!), whose name you've been trying to remember, had family. So you stop making the cheese sandwiches for a moment—you get
interrupted
by the
try-to-remember-boat-guy's-name
daemon that has just jumped at its chance opportunity—because you now
almost
have boat-man's name. . . .

Thus, these background processes, or daemons, are an important part of human cognitive processing. And, it seems that the more typically intellectual one's life is—I am guessing associated also with a more high-functioning-personality lifestyle—the more important these background daemons become. In concussives, these processing daemons are negatively affected, sometimes dramatically, as follows:

Because daemons use up resources, a healthy brain regulates triggering them: it generally won't spawn a daemon unless that daemon is likely to do something useful, and it will put a cap on the number of daemons it will run simultaneously. But even in healthy brains this triggering process is not an exact, or even very conscious, process. You might, for example, get a song stuck in your head, and be driven by curiosity to
play it over and over until you figure out what it is—even when you do not like the song, care about it, or care what its name is. A delicate balance exists between firing up daemons that might come up with something useful—following creative and heuristically intelligent “hunches”—and filtering out possible triggers that are not likely to lead to useful results.

This filtering process is an important part of intelligence, and of the efficient use of whatever native brainpower a person has at his disposal. If Gina fires up too few daemons, she'll be a dull, plodding sort of person who learns slowly, and only what is clearly directed by others. If she fires up just the right number, it will help her to be a witty, inquisitive, intuitive, creative sort of person who seems to make connections that others do not. If she fires up too many daemons, she will be a distracted, brain-fatigued, nonlinear, confused sort of person with all sorts of ideas that no one seems to “get,” and which do not lead to much.

In concussives, the filtering process itself is affected. Inappropriate daemons are continually triggered, needlessly searching for meaning in unfiltered minutiae. Concussives' automatic sensory filters no longer work correctly, so the world has a tendency to become a nightmare of cognitive input that is “noticed,” and thus must be
consciously
filtered: the sound of a truck driving down the road outside is given the same initial importance as the sound of a question from your daughter, sitting in front of you, with whom you are having a conversation:

Your daughter asks, “Hey Dad, have you seen my car keys?”

You are trying to parse the words, to separate them from the rest of the aural stream. A daemon has fired itself off and is looking for how the truck sound relates to the sound of the word
keys
: The truck is a
garbage
truck, which sounds like
garage
, which is where
cars
live. Cars have
keys
. This train of thought interferes with the understanding of your daughter's question.
Why would the keys be in the garage? Did I see her keys in the garage?

A concussive also loses the ability for his daemons to interrupt other daemons in the middle of their processing. I myself experienced this frequently: the phenomena of realizing that I knew something, and
knowing
that I knew it, but not being able to use that information, while at the same time realizing that the conscious, focused part of my brain was in need of exactly that same information.
*

Lastly—and very importantly
—
because low-level visual/spatial representations can be damaged in a concussive's brain, daemons may not realize, so to speak, that their conditions for termination have been met—the
match
against current circumstances fails—and they just keep running, well beyond their useful life.
*

In each of these cases—spawning unneeded and unwanted daemons, being unable to solve problems because of a lack of information even though it may actually have already been
located through another part of the brain's processing, and being unable to terminate daemons that have completed their tasks but can't form a visual/spatial pattern match to realize it—the result is that the brain grows increasingly fatigued, which in turn causes increasing difficulty with both sensory filtering and daemon communication, in a downward spiral of cognition failure.

From the outside, it just appears that a concussive is quirky and unreasonable about the noise of garbage trucks and the need for quiet in the household. He is slow in responding to simple questions about car keys, and asking such questions can make him unaccountably distressed. This does not make much sense to others, who are not aware of the processing overload that is going on under the hood.

You will often hear concussives complain that they “can only do one thing at a time.” And this is true. The sad fact, however, is that normal cognition, even when only doing one thing at a time on the surface, often requires many layers of simultaneous processing in the brain.

DAEMON GUILT.
Around this time, still in the early days after the accident, I had my first encounters with episodes of undifferentiated guilt: a guilt that crept up on me, and was triggered by subconscious processes, but which was not bound to any specific intentional actions I had taken. Such guilt feelings became quite common over the years.

Typically, I would be unsuccessfully trying to perform some task in my life—such as feeding myself during the apple-or-salami incident. I
knew
that I knew exactly how to perform the task, but I could not seem to access that knowledge of how to do it. At this point a “guilt daemon” would fire up, presumably
to get me to stop screwing around and conform to societal conventions of handling my own problems, instead of acting so helplessly: after all, as far as the daemon was concerned,
all the knowledge to get the job done was available to me.

I was left with this often-repeating circumstance where I knew how to act, knew what was to be done, felt guilty that I was not acting, but was powerless to initiate action toward my goals. But I was unable to make the guilt daemon “understand” that there was no way I could live up to the principle for which I was being held accountable. So, it would not go away, and actually made things worse by itself consuming precious resources.

Several years after the crash, Jake and I were leaving the Century Movie Theater in Evanston. Jake recognized that I was having some trouble walking—the parallel lines and moving planes on the long escalator ride down to the lobby were playing havoc with my highly vision-dependent balance systems. So, just as we were exiting through the glass doors of the theater, he said, “You can stay here if you want, and I'll go get the car.”

Because of
cognitive slowing
(which we'll discuss in detail in a later section), one symptom of which was my inability to turn spoken sentences into meaning quickly enough to keep up with normal speech, I did not quite make out what he said. After some error correction, based on the few words I did get, and the overall sound of his utterance, I mistakenly thought Jake had said, “You can stay here if you want, and I'll try to make out what those are.”

There were some pretty female Northwestern University grad students on the sidewalk to our left. As near as I could tell, Jake, an always-interested bachelor, was referring to the
attractive women, although I couldn't figure out what his comment meant. I wondered what my staying by the door of the theater had to do with Jake and the students.

While I was thinking this over, I crossed the street with Jake, slowly walking alongside him, following him back toward the car. I looked toward the women and said, “I'm sorry, Jake, could you explain what you mean? I don't know how to decide if I should wait while you figure out what something is.”

At exactly the time I started to speak, I also began to feel very distinctly guilty, which was unpleasant—a feeling of unnamed dread.

But this time, rather than deal with it on my own, I mentioned it to Jake. I was ready to look for some answers, and to analyze the details of what had just happened. Jake was game: scientific analysis of most any aspect of the natural world was something that he and I engaged in often. So we sat in his car and worked it out.

When Jake first spoke his sentence, I honestly did not hear all of the words clearly. This happens to all of us. Fortunately, however, most of us are almost instantaneously—and certainly without conscious thought—able to disambiguate an incorrectly heard utterance by replacing some words with others, so that the expressions make sense in the current circumstances: “. . . stay here if you want, and I will go get the car.” In computer science terms this is simply a matter of meeting the constraints of the words that we
did
hear and those of the context in which we heard them, while searching for possible candidate-phrases with which to fill in the blanks, and then ranking our proposed solutions for viability. At some point one of the solutions is deemed
good enough
and we abandon further search.

However, my own ability to perform this kind of
constraint-based searching for the purposes of verbal stream error correction had been compromised, which is why certain kinds of conversations could make me very tired, very quickly. Yet this was a matter of degree: my error-correction processing was mostly still intact—just moving slowly.

Thus by the time I had determined the need for further information, formed the sentence that could request this information from Jake, and constructed the sounds necessary for uttering the question, a different part of my brain had already figured out what Jake had actually said.

Under the hood, so to speak, two daemons,
S
(
Search
for meaning) and
Q
(ask
Question
), had been started up, followed later by a third,
G
(
Guilt
).
S
was an independent
search daemon
whose job it was to find out the meaning of Jake's utterance. Failing to retrieve enough information to disambiguate the sentence on its own,
S
activated a
query daemon
,
Q,
to ask Jake to explain. But in the meantime,
S
kept searching. Because
Q
involved the forming of an utterance—a spoken sentence—it became the center of my attentional focus. That is, I was intentionally asking Jake a question; thus, everything else receded into the background, including the ongoing search by
S
for the error-corrected meaning of what Jake had originally said to me. In the middle of actually asking Jake the question,
S
had simultaneously figured out what Jake had said—obviating the need to ask the question at all.