The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Catalyzing Inquiry at the Interface of Computing and Biology
They function with high degrees of autonomy. Some biological entities—such as neurons in a brain—can configure themselves automatically into networks (and reconfigure themselves to some degree when parts are damaged or destroyed). Sensory systems rapidly pick out salient features buried in large amounts of data. Many animals learn from their environment and become better adapted to what they are supposed to do. All biological organisms have mechanisms for self-repair, and all multicellular organisms grow from an initial state that is much less phenotypically complex than their final states.
Carver Mead once noted that “engineers would be foolish to ignore the lessons of a billion years of evolution.” The solutions that nature has evolved to difficult engineering problems are, in many cases, far beyond present-day engineering capability. For example, the human brain is not fast enough to process all of the raw sensory data detected by the optic or auditory nerves into meaningful information. To reduce processing load, the brain uses a strategy we know as “attention” that focuses on certain parts of the available information and discards other parts. Such a strategy might well be useful for an artificial machine processing a large visual field. Studies of the way in which humans limit their attention has led to computational models of the strategy of shifting attention. Such models of biological systems are worth studying even if they appear intuitively less capable than computation, if only for the fact that no machine systems exist that can function as autonomously as a housefly or an ant.
On the other hand, biological organisms operate within a set of constraints that may limit their suitability as sources of inspiration for computing. Perhaps the most important constraint is the fact that biological organisms emerge from natural selection and the evolutionary process. Because selection pressures are multidimensional, biological systems must be multifunctional. For example, a biological system may be able to move, but it has also evolved to be able to feed itself, to reproduce, and to defend itself. The list of desirable functions in a biological system is long, and successfully mimicking biology for one particular function requires the ability to separate nonrelevant parts of the system that do not contribute to the desired function. Furthermore, because biological systems are multifunctional, they cannot be optimized for any one function. That is, their design always represents a compromise between competing goals. Organisms must be adequately (rather than optimally) adapted to their environments. (The notion of “optimal design” is also somewhat problematic in the context of stochastic real-world environments.) Also, optimal adaptation to any one environment is likely to disadvantage an organism in a significantly different environment, and so adequately adapted organisms tend to be more robust across a range of environments.
The evolutionary process constrains biological solutions as well. For example, biological systems inevitably include vestiges of genetic products and organs that are irrelevant to the organism in its current existence. Thus, biological adaptation to a given environment depends not only on the circumstances of the environment but also on its entire evolutionary history—a fact that may well obscure the fundamental mechanisms and principles in play that are relevant to the specific environment of interest. (This point is a specific instantiation of a more general phenomenon, which is that our understanding of biological phenomena will often be inadequate to provide detailed guidance in engineering a computational device or artifact.)
A corollary notion is that nature may evolve different biological mechanisms to solve a given problem. All of these mechanisms may enable the organism to survive and even to prosper in its environment, but it is far from clear how well these mechanisms work relative to one another.2 Thus, which one of many biological instantiations is the most appropriate model to mimic remains an important question.
Finally, there are only a few examples of successful biologically inspired computing innovations. Thus, the jury is still out on the ultimate value of biology for computing. Rather than biology being helpful across the board to all of computing, the committee believes that biology’s primary relevance (at least in the short term) is likely to be to specific problem areas within computing that are poorly
For example, fish and squid use different mechanisms to propel themselves through the water. Which mechanism is better under what circumstances and for what engineered artifacts is a question for research to answer.