I have been thinking about the idea of the extended mind for a couple of years now, and the design of tools as essentially parts of the human mind. Its from this perspective that I really love the work of Bret Victor – especially his ‘Media for thinking the unthinkable’. It was therefore with great interest that I first heard David Krakauer talk about his ideas about ‘cognitive artifacts’ (I think this term was coined by cognitive scientist and design guru Don Norman), on Sam Harris’s podcast. David Krakauer is a complexity scientist and evolutionary biologist who heads up the Santa Fe institute, and sounds like a fascinating guy – I recommend listening to the podcast if you would like to hear more from him. A lot of the discussion of cognitive artifacts, and Krakauer’s interest in them currently is in the context of the threat of AI, and whilst I think his specific concerns have some validity, I want to rather focus on how thinking about things in his terms can be useful for design.
Cognitive artifacts are essentially tools we have created to help us think – that in some way ‘bolster’ our intelligence. An abacus is a prime example: a physical tool that allows us to achieve a higher level of arithmetical competence by just relying on more basic know-how, and the functionality of the tool itself. However, things get a lot more fundamental and subtle than that, because the key feature of cognitive artifacts is really that we internalise them (cognitively), and they then become parts of our minds, that we can use to think in new ways without using the tool anymore.
Krakauer gives the example of mathematical reasoning itself – this is not a skill one is born with, but rather one learns the cultural knowledge of numbers and rules that when correctly internalised (after years of learning the ‘technologies’ of maths with a pen and paper and so on), results in mathematical reasoning ability. There have been different number technologies, such as Roman numerals, but literally by virtue of the rules governing how they are put together and combined, the system was not flexible and powerful enough to breed the wealth of mathematics we have today. The ‘machinery’ of our (originally Indian and Arabic) number system has such powerful properties that we are able to internalise the technologies relatively easily, and the result is a new ‘internal’ cognitive ability.
And this gets to the root of the difference between what Krakauer calls ‘complementary cognitive artifacts’ vs. ‘competitive cognitive artifacts’. Complementary ones are like our number system, and to a slightly lesser extent the abacus. Artifacts that afford us new cognitive abilities when we use them, but which also allow us to learn and internalise these abilities, such that we can eventually discard the tool and work with just a virtual version – a mental representation of the tool. Compare these types of cognitive artifacts with a modern calculator: a tool which gives us far superior cognitive abilities than an abacus whilst we are using it, but leaves us with no enhancements when discarded. In fact, many argue it leaves us less enhanced, less able, because of our learned habit to offload thinking to the tool, with no complementary learning of how the tool does its job. These are the ‘competitive’ cognitive artifacts, since in some way they compete with our own abilities, rather than complementing them.
Competent abacus users can eventually perform more complex mathematical processes just using their minds because they can picture using the abacus in their mind’s eye (or just as likely also ‘feel’ using the abacus with their fingers even when it isn’t there). The mechanism of the technology is exposed and simple enough to be reliably internalised as a mental representation, or simulation. Once the artifact is adequately internalised, it can contribute to ‘real’ thought in a very immediate and seamless sense. The same can never be true of a modern calculator, since we firstly have no way to intuitively grasp how it is doing its work, and secondly the abilities of the calculator far outstrip the natural potential of our own in many ways.
Of course this is not a binary, black and white distinction. No cognitive artifact is completely complementary – there are still many operations that even the most experienced abacus master in the world will not be able to perform without the abacus, and of course most of us would still need to write down difficult multiplication and long division problems to work through them. On the other end of the spectrum, the extent to which the effects extremely competitive cognitive artifacts are detrimental is debatable. Krakauer himself is worried that AI will be the killer competitive cognitive artifact, that we begin to rely on intuitively and effortlessly for a wide range of cognitive tasks, gradually forgetting (both personally and culturally) many fundamental cognitive abilities. Others might argue that our growing integration with artificial intelligence will lead to new forms of thought, and new cognitive abilities that we cannot even imagine yet. What seems undeniable however is that we will not be able to internalise many of thee abilities in a way that would have value without the AI, since the nature of what the AI would be providing to us would almost by definition be far too complex for a single human mind to understand – that’s what we have the AI for.
No matter the outcome of the debate around competitive cognitive artifacts, I think its a great mental model to have in mind when designing tools of any kind. The more powerful and useful tool is one that has some degree of being a complementary cognitive artifact. It is something that opens someone’s mind to a new way of thinking, that then stays with them forever. It is probably a good exercise in diminishing one’s ego as a designer to try and design tools that have the very purpose of becoming obsolete one day, not because they break, but because they have given the user the ability to achieve whatever it was they needed the tool for, without the tool at all anymore.