I regard the Southwestern desert of the United States as the most fiercely beautiful real estate on earth. Many human cultures have passed through, vied for and settled in this magnificence, but most have rested lightly (with a few exceptions, most harshly the bright green lawns of Anglo mansions, for no vegetation could be less appropriate to the landscape or more wasteful of precious water in the desert). Seen from a distance, Hopi towns on the mesa tops look like horizontal layers of sedimentary rock, with ladders protruding from kivas as the only clear evidence of human construction for the uppermost stratum. Science has also entered this region of maximal diversity by building institutions as varied in purpose as Los Alamos, for the superior prosecution of war, and the Santa Fe Institute, for better understanding of complexity.

George Johnson, until recently an editor at The New York Times, has lived in the various scientific subcultures of the region, and has studied the primary local alternatives of American Indian, Hispanic and Anglo origin. He uses this maximal diversity to ask the most important of all questions about the power and social impact of science: "Do the patterns found by the scientific subcultures of Santa Fe and Los Alamos hold some claim to universal truth, or would a visitor from a distant galaxy consider them as culturally determined as those divined by the Tewa and Penitentes?"

As with all oversimplified dichotomies, as Mr. Johnson knows and acknowledges, both sides contribute essential pieces to a full puzzle of many more dimensions. In the oldest of conventional arguments (none the worse for wear), science must have some privileged position in its area of stated competence -- the description and explanation of empirical reality, but not the exploration of moral or esthetic truth -- because the technological fruits of its application so clearly work (whatever the consequences for our lives). But all science, necessarily done by human beings with psychological hopes and cultural expectations, must be socially embedded, and must therefore reflect the changing mores of historical moments.

Such social biases have been much studied and long understood by historians of science. Mr. Johnson focuses this excellent book on another species of bias less often treated by critics but forcefully raised by the sciences most closely examined in this book: those that operate at sizes and scales most distant from our ordinary experience and perceptions, including cosmology, particle physics and quantum theory.

The human brain is an evolutionary instrument designed to deal with our direct experiences and dangers: times in seconds to years, sizes from gnats to elephants, and speeds from inchworms to cheetahs. Why should we be good at protons and galaxies, not to mention infinity and eternity?

Yet we are driven to wrest order and meaning from nature's mixture of complexity and randomness. We have always constructed elaborate theories of order, yet we know that the smartest scientists of past centuries accepted notions now identified as ludicrously inadequate, including a small universe centered on the Earth, and a created history of life extending back only 6,000 years. Surely our current theories lie closer to empirical adequacy. But we may still be light years from accuracy, particularly on theories built on cascades of assumptions, if such a notion as final accuracy even has meaning beyond our hopes and neurological needs.

We are particularly bad at a clump of related concepts vital to understanding complex systems: randomness (for our explanations are stories, and we prefer to tell tales about conventional causes and intents); contingency (for we yearn to depict our complex historical present as a necessary outcome of former conditions, and not as an unpredictable, if sensible, result that would never occur again if we could replay life's tape from an identical starting point); and hierarchy (for we relish the simplicity of applying a set of causal principles across all levels of size and time, and encounter immense mental difficulty in grasping, as Mr. Johnson states, "the notion of rules on one level giving rise to completely different rules on another level").

Mr. Johnson shows admirable openness and agnosticism as he examines, chapter by chapter, a range of scientific frontiers furthest from the simple and deterministic physics and chemistry of objects operating at our scales and times. But he occasionally shows the hand of conventional bias, thus affirming by example (and perhaps unintentionally) the mental obstacles that stand against our restructuring of nature.

He is particularly reluctant to view the history of life as undetermined in broad outline and not predictably driven to greater complexity (although he gives my own contrary views both a lucid explanation and a fair summary). And he honestly locates this reluctance in our social traditions and psychological preferences. He writes, "Most of us feel certain that the biological world inexorably increases in complexity." But why do you feel this way, Mr. Johnson, in a world that has been truly dominated by bacteria since life began?

I certainly share Mr. Johnson's welcome skepticism in treating sciences of unfamiliar scales, times and speeds, where apparently "firm" conclusions rest upon long cascades of interacting assumptions. (More cynical folk might make an analogy to a house of cards.) I cannot be persuaded to fight and die for a cosmology that places 90 percent of cosmic matter into the category of "currently unobservable."

Advances in instrumentation have spurred the growth of science by giving us access to previously unknowable scales. But the human mind acts as an ultimate limit and gatekeeper, whatever its magnificence as an instrument of understanding. In the "Essay on Man," Alexander Pope questioned the utility of better perception into invisible scales, fearing that we might lose philosophical "sight" of grander things:

Say what the use, were finer optics given:
To inspect a mite, not comprehend the heavens?

But how much of the heavens can we comprehend without trying to view all of nature's scales? And how does our gloriously constrained brain suffer from its evolutionary derivation for problems in one small corner of reality?