Input, Output or Throughput?

We live in an input-output Universe

Or so we believe.

Pretty soon, if we survive, everything will give us feedback. The Cognitive Internet of Things will combine IoT devices with artificial intelligence to tell us just exactly who and what we are- according to robots.

One paper I read recently has -nearly -the best title of any academic paper I have ever read.

Who could not resist “Nuclear Norm Regularized Structural Orthogonal Procrustes Regression for Face Hallucination with Pose” by Dong Zhu, Guangwei Gao, Hao Gao and Huimin Lu? So cool.

Feedback is, of course, incredibly useful. Sometimes in ways that we don’t even notice. And the way we develop our responses to feedback involves a range of cognitive functions such as foresight, range selection, organisation through ranking, and/or rating, choice, and so on. We also require the cognitive ability to combine foresight and insights to respond to changes in either on the intrinsic materials available for making tools or on the extrinsic conditions for their use.

There is archaeological evidence that in Africa about 3.4 million years ago stones were being flaked by australopithecines -bipedal hominoids- who used the resulting sharp edges in ways that left cut-markings on animal bones, presumably as a result of attempting to scrape edible meat and fat from the bones of dead mammals.

If this interpretation of the evidence is correct, then stone tools were sometimes made and used to process animal remains for food- butchery, if you like. This raises the question of whether or not Australopithecus (and their human descendants) could hold -simultaneously in mind- combinations or permutations of ways to prepare stone artifacts with their possible future uses as tools appropriate for particular purposes.

And they also had to hold those models in the face of a continuously changing shape as the stone lost flakes.

This is the essence of predictive model making.

Of course, there are different approaches to making models.

The first is connatural- it engages with some type of external reality. The second is conceptual- it engages with other concepts describing that reality with greater or lesser accuracy, completeness, etc.

But one of the things people 4 million years ago included in their models was themselves. They were deeply embedded in the physical reality they occupied. Their activity was a part of the system they tried to explain

Now we are much more conceptually driven. We see our models as being ‘outside’ ourselves. This allows us to generalize- at a price. If we don’t ‘own’ the models we develop, we become less engaged with them. And they must become less effective. Because, of course the anthropic principle, and its relationship with the cosmological constant, tells us we do not stand outside the Universe. We are, in fact part of the Universe.

And our models- not our products- need to give us accurate feedback not just about how they work.

But, also about how we work,

If there is to be any useful evolution of the way we work mentally, physically and emotionally.