As one learns more about neuroscience and those who study it, there is almost a sense of Greek tragedy — an element of hubris is always lurking about. What bravado to study something so complex and so …well … incomprehensible! What chutzpah to talk confidently about what we know in spite of the vastness of our ignorance! In recent years, a seemingly overt expression of this attitude has surfaced in outrageous book titles. Have the authors of How the Mind Works or Consciousness Explained totally lost touch with reality? I prefer to think that these titles are tongue-in-cheek and not to be taken too seriously. Substantial works that would merit such titles are probably beyond the reach of our lifetimes, and the authors presumably suspect that their readers know this before picking up these books, though I am less certain about their marketeers.
But neuroscientists do study these systems, so they must believe that eventually some significant understanding will be achieved. Is this hubris, or not? One possible answer is that progress is being made, so why not continue? A more satisfying answer is that some of the research that has been done on the brain and its origins suggests that the brain holds within itself properties that will lead to its eventual understanding.
Simplification by Abstraction
Before reviewing the evidence for these properties, consider something much simpler to understand, the computer. The basic component of the computer is the transistor. In theory, if you understand a transistor and some other basic electrical components, then you have the foundation for building a computer from scratch. However, if everyone who worked with computers to produce the movie “A Bug’s Life” had been forced to envision the tasks that they needed to accomplish in terms of the intimate workings of a transistor, it would have been the world’s shortest movie. Instead, engineers take the relatively complex device, constrain its operation to a small working range of carefully controlled parameters, and thereby create an abstract device with relatively simple properties. In this way, transistors with their nonlinear electrical properties are converted into logic gates, which can be thought of abstractly as converting true and false inputs into true and false outputs in well-defined ways. This process is repeated to create increasingly complex hardware, and an analogous process is used to create complex software. The components at one level of abstraction serve as the building blocks for creating the next level of complexity.
This trick works well for devices that are designed to be components in a larger structure, such as electronic chips in a computer, but what of understanding biology, and in particular, the brain? Certainly, fields of science are divided as if we can understand them within the context of their own abstracted components. We don’t expect all scientists to explain their findings in terms of quantum mechanics. Are these lowered expectations due to our acceptance of our limited mental capacity, or are there aspects of the physical world that warrant this approach? The fact that many scientific fields are able to create explanations and accurate predictions without relying on a quantum mechanical basis suggests that this approach is not without merit. The key to these successes is the ongoing development of meaningful abstractions on which to base the underlying understanding. This development begins by considering the constraints that these abstractions must meet.
Genetic Support of Simplifying Principles
A single cell, namely the zygote, formed when a sperm delivers its genetic contents to an ovum, contains the information for creating a brain within the embryo in the sheltered environment of the womb. Most of this information is thought to be contained within the genetic code. This code is, however, dwarfed by the complexity of the brain. The genetic code is estimated to have about 100,000 genes, but there are 100,000,000,000 neurons with as many as 1,000,000,000,000,000 connections between them.
How can such complexity be generated from such a small code? We must assume that there are general patterns that are repeatedly used throughout the brain, thus reducing the amount of information required for their specification. The number of patterns must be significantly smaller than the number of genes, which implies that these patterns, once discovered, might provide the required abstractions from which to create a coherent model of brain function. These patterns are likely to exist at many levels of organization of neural tissue, so we must search for them as canonical descriptions of neuronal types, circuits, maps, and cognitive systems. The limited number of genes suggests that these descriptions do exist.
Evolutionary Support of Simplifying Principles
Evidence from human evolution also suggests that canonical descriptions exist. A Homo erectus specimen that was dated as living 1.6 million years ago had a brain estimated to be about 900 cc, as compared to a 1350 cc brain of an average human today. Doing some back of the envelope calculations, we can estimate that we have obtained, on average, about 30,000 neurons per generation. Obviously, it would be impossible for random mutations to explicitly specify the wiring of so many new neurons each generation. Since most of this expansion occurred within the neocortex, we can assume that the neocortex contains circuitry that is reused over and over again. Furthermore, we can assume that this circuitry is so generic, that benefit was conferred to early humans by simply adding more of it to the cortex.
Physiological Support of Simplifying Principles
Canonical descriptions are also supported by physiological evidence. Cortical plasticity is a phenomenon that has been studied in the sensory and motor systems of mammals. In these studies, it has been demonstrated that an area of cortex, representing a particular portion of the sensory or motor field, can change its function in response to repeated performance of a task that is important to the animal (i.e. allows the animal to obtain food). Experiments in humans have likewise shown activation of visual cortex in blind subjects reading Braille. This ability to adapt the cortex to the required set of tasks by rearranging the cortical resources devoted to each task implies that the underlying cortical circuitry is generic and largely independent of the specific task being performed.
The Path Ahead
The genetic, evolutionary, and physiological evidence all suggests that though the brain is complex, it must contain some general principles within its specification. We can use these principles to simplify the task of understanding the brain as they will provide us with adequate abstractions with which to describe it. Does this argument take us outside the realm of hubris? Only if we are patient and farsighted. We can not delude ourselves into believing that we know more than we do. We must continue to struggle to achieve the understanding that only our descendents may know. We must prepare for the Century of the Brain.