|
Artificial neural networks patterned on fundamental neurological features of the human perceptual system have been shown to produce Bayesian probabilistic classifications of galaxies1-3, identify biotic and abiotic alteration of subsurface basalts4, distinguish terrestrial fossils from their background rock matrix5, and detect areas of Archean hydrothermal alteration6. Data inputs for these classification tasks have varied from astronomical or high altitude
images and spectra, to sub-micron resolution elemental abundances. However, Bayesian theory assumes an absence of statistical and interpretive ambiguity in a target signal, the antithesis of the problems facing remote and human exploration of extreme environments on Earth and extraterrestrial sites such as Mars, comets, and the icy moons of Jupiter and Saturn. Fundamental to our certainty about the classification of geobiological targets on Earth is a long scientific history of familiarization both with the geochemical evolution of our planet and the reliability and discriminating power of particular instruments. Reduction of the uncertainty associated with a putative extraterrestrial biosignature derived from a single probe is most often attempted by deploying a suite of instruments, each one interrogating distinct morphological and chemical phenomena in a target7. But understanding the relative weighting appropriate for merging disparate signals or distinct data sets is not a trivial issue8. And, as we have most recently seen in the case of ALH84001, strategies relying on the cumulative statistical power of multiple probes often crumble when subsequent review of abiotic physicochemical phenomena reveals even a single abiotic mechanism, no matter how improbable, capable of replicating the putative biotic signal. Finally, for extend extraterrestrial missions or work in remote environments on Earth, the fundamental "fewest moving parts" reliability rule must come into play. This communication highlights the minimum requirements for an astrobiological instrument suite for remote or human exploration of extreme environments both here on Earth and in our local and neighboring planetary systems. Critical items of concern include obtaining co-registered data characterizing target morphology, metabolism, and mobility; the face validity and familiarity of the instrumentation to the scientific community, and the choice of instrumentation sufficiently inexpensive and easy to use that it might find wide spread usage within the astrobiology community prior to mission deployment. Preliminary indications are that such an instrument can be implemented for a cost accessible to high school, college, and graduate students interested in geobiological and astrobiological research in extreme or hazardous environments.
|
