The extent to which evolution incorporates randomness has been debated by scientists throughout history, but a new paper claims the phenomenon may be more 'intelligent' than classical theory acknowledges – and not due to a theological perspective.

According to Richard Watson, an evolutionary biologist at the University of Southampton in the UK, evolution can 'learn' from previous experience, which could help explain how a process of random, natural selection can seem to produce such intelligent designs.

In an opinion piece in Trends in Ecology & Evolution, Watson and co-author Eörs Szathmáry from the Parmenides Foundation in Germany suggest that we can still improve our understanding of how evolution works. We can do this, they say, by encompassing the way learning systems like neural networks develop intelligent behaviours based on previous experience.

"Darwin's theory of evolution describes the driving process, but learning theory is not just a different way of describing what Darwin already told us," said Watson. "It expands what we think evolution is capable of. It shows that natural selection is sufficient to produce significant features of intelligent problem-solving."

Conventional evolutionary theory describes how random variation and selection is sufficient to provide incremental adaptation as organisms evolve over time. But according to Watson and Szathmáry, this system isn't incompatible with learning theories that demonstrate how incremental adaptation (such as in a neural network) is sufficient for a system to exhibit what we would call 'intelligent' behaviour.

"When we look at the amazing, apparently intelligent designs that evolution produces, it takes some imagination to understand how random variation and selection produced them," said Watson. "But can natural selection explain the suitability of its own processes? That self-referential notion is troubling to conventional evolutionary theory – but easy in learning theory."

The authors say new kinds of research are already demonstrating this kind of learning evolution, including selection in asexual and sexual populations with Bayesian learning and the evolution of ecological relationships with distributed memory models.

"Learning theory enables us to formalise how evolution changes its own processes over evolutionary time," said Watson. "For example, by evolving the organisation of development that controls variation, the organisation of ecological interactions that control selection or the structure of reproductive relationships that control inheritance – natural selection can change its own ability to evolve."

If scientists can encompass this kind of learning theory into how they study evolution, the authors say we'll be able to solve evolutionary puzzles in new ways and enjoy greater insight into how evolution produces its amazing results.

"If evolution can learn from experience, and thus improve its own ability to evolve over time, this can demystify the awesomeness of the designs that evolution produces," said Watson. "Natural selection can accumulate knowledge that enables it to evolve smarter. That's exciting because it explains why biological design appears to be so intelligent."