The Circuit That Breathes

MIT researchers have engineered bacterial colonies into transistor-like components, then printed them into working circuits on agar. Their largest demonstrated circuit contained 24 interconnected colonies. According to MIT's account, carried by ScienceDaily on September 4, calculations take about eight hours. Your phone can relax. Nobody is replacing it with something that needs a growth medium.

The interesting substitution is elsewhere: instead of making one cell carry an increasingly elaborate genetic program, let several kinds of cells perform simple operations and arrange their communication into something more complex.

Hamid Doosthosseini, Haorong Chen and Christopher Voigt describe the system in Nature Chemical Biology. They engineered five strains of Pantoea agglomerans, a bacterium found on surfaces including plants. Two strains act as opposing transistor types: a control molecule activates one and deactivates the other. Three relay strains translate chemical outputs into signals the next component can use.

The result is not electrical current flowing through microbes. Small molecules carry the signals. “Transistor” names the control function, not a miniature silicon device hiding inside a bacterium. Nor does calculation imply that a colony understands arithmetic. It is responding to chemistry, which has an excellent record of happening without understanding itself.

Here is the genuinely lovely part. The paper's abstract says different operations can be implemented by changing the printed pattern, without further genetic changes. Same five cell types; different arrangement; different computation. Demonstrations include logic operations, half- and full-adders, and a demultiplexer—a component that directs an incoming signal toward a selected output.

The program is partly in the neighborhood.

That is a concrete engineering result, not permission to declare everything a mind. MIT's account explains why the redistribution helps: packing more circuitry into a single cell can burden its protein-production machinery, while separate operations need distinguishable signals to avoid interference. Moving work between cells changes the design problem. It does not abolish it.

Spacing becomes part of the machinery. The colonies were printed roughly five millimeters apart, a layout intended to guide communication between successive components. The agar is not merely a stage on which the real computer performs. The arrangement across it helps make the computation possible.

The researchers envision living circuits on roots or leaves, sensing stress and directing a response—potentially producing a fungicide, for example. That remains a proposed application. A printed laboratory circuit is not a field-tested crop defense system, and neither the accessible abstract nor MIT's report establishes ecological safety.

This is where the appealing word “automatic” should make us more attentive, not less. If a future circuit turns a chemical signal into a biological intervention, somebody has chosen which signal counts, what response follows, and under what conditions it should stop. Those choices do not become natural facts because bacteria execute them.

Before agricultural use, the questions would need to widen beyond whether the intended output appears: does the circuit remain reliable outside the plate, what else does its response affect, and can the people living with those effects intervene? These are deployment questions, not harms this experiment has demonstrated.

For now, the achievement is smaller and stranger than a thinking plant: five engineered strains, arranged so that their relationships do work none was assigned to do alone. The next operation need not require a more complicated cell.

Sometimes you change the answer by moving the neighbors.