How technology can improve without increased understanding

Archeologists sometimes find ancient artifacts that show considerable sophistication. These might cause us to infer that the people of that time had a better understanding of the underlying science than we had previously given them credit for. This is because advances in technology often go hand in hand with advances in science. Technological advances upon up new frontiers for scientific investigation while new scientific theories lead to new technologies. But that link may not always exist.

This article discusses an experiment where successive generations of students were able to steadily improve a product without gaining a corresponding understanding of why the product worked better.

Researchers used “chains” of volunteers to tackle an engineering problem, with each volunteer able to learn from the last. Solutions improved with each “generation” — but those at the end of the chain had no more understanding of key concepts than their predecessors.

“We tend to explain the existence of complex technologies by saying humans have big brains and superior causal reasoning abilities,” said Dr Maxime Derex, of the University of Exeter and the Catholic University of Lille.

“But — as our study shows — you don’t have to understand how something works in order to improve it.

“Artefacts from hundreds or thousands of years ago do not necessarily show that their makers had a plan or a theory about how something would work.”

The researchers said: “Most participants actually produced incorrect or incomplete theories despite the relative simplicity of the physical system.”

The findings prove the power of “cultural transmission, without the need for an accurate causal understanding of the system,” they said.

The journal article (paywall) is here and its abstract states:

Bows and arrows, houses and kayaks are just a few examples of the highly optimized tools that humans have produced and used to colonize new environments. Because there is much evidence that humans’ cognitive abilities are unparalleled, many believe that such technologies resulted from our superior causal reasoning abilities. However, others have stressed that the high dimensionality of human technologies makes them very difficult to understand causally. Instead, they argue that optimized technologies emerge through the retention of small improvements across generations without requiring understanding of how these technologies work. Here we show that a physical artefact becomes progressively optimized across generations of social learners in the absence of explicit causal understanding. Moreover, we find that the transmission of causal models across generations has no noticeable effect on the pace of cultural evolution. The reason is that participants do not spontaneously create multidimensional causal theories but, instead, mainly produce simplistic models related to a salient dimension. Finally, we show that the transmission of these inaccurate theories constrains learners’ exploration and has downstream effects on their understanding. These results indicate that complex technologies need not result from enhanced causal reasoning but, instead, can emerge from the accumulation of improvements made across generations.

I recently attended a talk and demonstration about Stone Age technology used in Ohio about 13,500 years ago. One of the items shown was a spear that could be thrown with the very high speeds needed to bring down large animals. The speeds needed were greater than could be obtained by simply throwing the spear. The early people had added an extra element that essentially added a slingshot mechanism to achieve the required speed and I was impressed at the idea behind the device and its elegant implementation. But it seems like they might have been able to arrive at it by piecemeal improvements without necessarily understanding the underlying physical principles.

This purely technological mode of development has the negative effect, as the authors of the journal article state, of constraining the range of exploration and limiting their later understanding. There is a rough analogy with evolution by natural selection. It proceeds by incremental improvements, not because of some comprehensive plan. I read an account about how evolutionary biologist Steven Jones once worked to design nozzles that would emit a better spray of liquid soap. By starting out with the design that worked best out of the ones they already had, they made small changes to it to find one those that worked better. From those, they again selected the best and made as series of small changes to it. And so on. They finally arrived at a much better nozzle than the one they started with without having had to use any basic scientific knowledge of how spray nozzles worked. But of course, the final product depended on the choices made at each step. A different, sub-optimal choice at any stage may have resulted in a superior product at the end but in the absence of an underlying scientific theory, that was not attained.

Evolution is also like that. There is no reason to think that the organisms we currently have, including humans, are the best possible result of the process because we were never part of some grand plan but instead are the end product of a series of piecemeal improvements. In fact, it is this particular feature of evolution that so arouses the ire of religious people for whom it is important that human beings be the result of a divine plan.


  1. says

    I often remind myself that the ancients weren’t stupid -- they were just low tech. Lately I have been experimenting with making my own bronze and casting it using lost wax process. It’s practically prehistoric, but when you look at the molds that ancient smiths used for casting bronze swords, you can see that they encountered the same problems (because physics hasn’t changed) that we still wrestle with today.

    What they did not have, because they did not need it, was precision. You don’t need metal lathes until you need steam engines, but you can’t have steam engines without them.

    Thy physics of tossing a spear are really interesting -- ditto a bow: the longer the path your missile takes while accellerating, the faster it will go -- therefore a longer/taller bow will hit harder, a rock thrown with a sling, and a spear thrown with an atlatl will hit harder, etc.

    There’s a really cool book I read a few years ago Brunnelschi’s Dome about the construction of the dome in Florence, starting in 1420 or so. They developed cranes, and ox-powered lifts capable of hoisting multi-ton rocks, and architected how to construct a dome from the inside, capping it at the point where the stones were nearly horizontal. They had to deal with expansion and heat wear in hemp ropes -- problems solved by using steel cable -- but overall, not much different from today.

    Watching an expert use an atlatl is a fun thing

  2. Sophy Cooper says

    Food preparation is another entire universe of technical improvements made often without knowing why something works. Leave the fruit out to get a little rotten and funky and, woohoo! does it make you feel good after you eat it. Then generations of experimentation with fermentation and we get delicious wines. Leave some barley and water out and evetually you get beer.
    Today many people follow recipes and bake wonderful things without knowing exactly how the ingredients work together, but will make adjustments for flavour and texture (or substitute because they ran out of something) until it doesn’t work or they’ve created something different.

  3. tenine says

    Another food technology that seem puzzling is the Native American Indians’ use of lye to treat corn (maize). This treatment improves the nutritional quality of the grain and was used in a number of societies; in the area I am most familiar withe, SE North America in the form of hominy. Did someone figure out that it was more nutritional by some experimental process? Was it just done just for taste and was only coincidentally superior in nutritional quality?

  4. Owlmirror says

    There is a rough analogy with evolution by natural selection. It proceeds by incremental improvements, not because of some comprehensive plan. I read an account about how evolutionary biologist Steven Jones once worked to design nozzles that would emit a better spray of liquid soap.

    The canonical example of weird solutions reached by evolutionary algorithms is Adrian Thompson’s work evolving FPGA circuits (lay description, also original paper).

    Taken from the first link:

    Five individual logic cells were functionally disconnected from the rest— with no pathways that would allow them to influence the output— yet when the researcher disabled any one of them the chip lost its ability to discriminate the tones. Furthermore, the final program did not work reliably when it was loaded onto other FPGAs of the same type.

    As noted, the end result was something that was uniquely idiosyncratic to that particular chip. Another issue was one of temperature — since the temperature had been held pretty constant for the entire process, the chips did not function the same at different temperatures (something the researcher considered to be expected). A suggestion for further experimentation would be to evolve in parallel in groups of chips held at different temperatures, only selecting those that worked in all of them to go forward to the next generation.

  5. Owlmirror says

    the longer the path your missile takes while accellerating, the faster it will go — therefore a longer/taller bow will hit harder

    Isn’t the composite bow a counterexample to the point of “longer/taller will hit harder”? That is, the bow isn’t longer or taller, but the additional flex in the bow allows the path to be made longer (and thus hit harder) than a bow of the same size that is not composite.

Leave a Reply

Your email address will not be published. Required fields are marked *