Books et al.Cell Biology

What Is It Like to Be a Cell?

+ See all authors and affiliations

Science  21 Aug 2009:
Vol. 325, Issue 5943, pp. 948
DOI: 10.1126/science.1176764

Nobody knows where consciousness comes from. In his provocative Wetware, cell biologist Dennis Bray presents the view that many features of conscious beings, including learning, knowledge, and awareness, are present within single cells.

Bray lays his foundation by describing the remarkable behaviors of single-celled organisms. He describes how the giant ciliate Stentor can not only run through a series of different strategies to avoid a noxious stimulus but also even learn over time which strategies work and which don't. His other examples include motile bacteria determining the direction of chemical gradients and amoebae actively pursuing swimming prey. He goes on to describe more-complicated cooperative behaviors that groups of cells can perform when they communicate and act together. His vivid descriptions paint a clear picture of cells as intelligent agents, continuously sensing their environments, computing the appropriate response, and learning from past mistakes.

At some level, Bray's argument is that cells are aware because they look aware. Although strictly speaking this is a non sequitur, on the other hand there is strong tradition in psychology of describing all cognition in terms of behaviors. Bray's introduction fits within this behaviorist tradition. But he then moves beyond behaviorism by describing how the biochemical systems that underlie these behaviors resemble neural networks capable of computation and learning. His consideration of computational mechanisms within cells sets a clear agenda for future research into mechanisms of cellular cognition.

Discussions of cognition often degenerate into arguments about terminology. Terms like “knowledge” or “learning” can mean different things to different people. Bray has a clear vision of what these terms mean to him, and he spells them out quite clearly. Nonetheless, sometimes the reader may need to consider whether Bray's use of such terms really matches our intuitive usage. For example, he argues that biochemical pathways are able to “learn” over evolutionary time in the sense that networks that do not perform the correct computation fail to be inherited and that therefore, at the end of the day, the networks implicitly represent some form of “knowledge.” Is that really what we mean by “learning”? Suppose we took a thousand horses, asked them what is one plus one, and then shot all those that didn't stomp their hooves twice. Nobody would say that the horses learned how to add. This clearly represents a confusion of levels, in that the unit of learning would be the ensemble of horses, not an individual horse.

CREDIT: COURTESY DENNIS BRAY

What makes a neural network capable of learning isn't the fact that we can kill off all the incorrect networks from some larger ensemble. Rather it is the way the elements and connection weights of a particular existing network can change their function in response to information about the performance of the network. This responsiveness requires plasticity within the system. There is a case to be made that biochemical networks may have such plasticity. Bray touches on this, but in general he shies away from this type of mechanism in favor of a purely evolution-based view of learning. Although the evolution of networks is no doubt very interesting, to say that a cell can learn because its biochemical networks can evolve over many generations seems to deviate from at least the standard usage of “learning.”

One of the book's major strengths is Bray's overall unity of vision; another is the way he marshals a breathtaking diversity of fields to make his case. For example, he offers entire chapters on biomimetic robots and synthetic biology. This breadth guarantees that almost every reader, regardless of prior expertise, will learn something new. Generally written at a very accessible level, the book clearly is targeted at the educated lay audience. There is probably a need for a more-technical presentation of Bray's computational view, one that would delve into more details. Although Bray does not attempt that in Wetware, he does point out a number of fascinating technical issues that could provide agendas for entire research labs. For example, he brings up the fact, often missed in discussions of synthetic biology, that mixing of diffusible molecules limits the computational capacity of a system with a fixed number of molecular components, whereas such limitations would be relaxed if the system were not well-mixed. A book written for researchers could provide whole chapters on anomalous diffusion, solid-state biochemical reactions, and so forth. Nevertheless, Bray has already done a great service in making these types of problems explicit.

And that is really the point of Wetware: to inspire us to reconsider our basic assumptions about what cells can do. However we choose to interpret words like “think” or “learn,” it seems safe to say that most biologists would hold that a cell cannot think, in any sense of the word. After considering Bray's arguments, they will probably be less dismissive of the idea. Whether cells think or not, there is no question that Wetware will get the reader thinking.

Related Content

Subjects

Navigate This Article