I wonder if you understand how evolutionary processes generate novel forms. If you do then think about a brain employing such a process of randomised combination, variation and selection to generate new ideas.jamest wrote:C'mon, that's just gobbledygook. Do you really think that there's any solid explanation within this statement?GrahamH wrote:If brains contain representations of objects, relationships, generally 'concepts', then new concepts can be generated by semi-random permutation combined with a selection process that might somehow recognise a new combination as a fit with a problem, task to be performed, opportunity etc.
Of course new ideas arise, but what is it that is new about them? Is it the elements, or the way the fit together?jamest wrote:Hold on. If new ideas/concepts don't occur to us, then how did "existing concepts" ever come into understanding?Entirely new ideas do not seem to occur to us. Invention takes existing concepts together with
The point is that each and every idea/concept must have an origin (they must have been 'new' at some point). Which means, undoubtedly, that the human mind has the capacity to create new ideas/concepts.
I think this is really a topic in it own right, but the premise is that concepts are recognisers of patterns. Neural networks are demonstrably recognisers of patterns and have learning ability. A child looks at the sun and a response forms in her brain. That response is the beginning of a concept of the Sun. The connections that respond when she sees the Sun link to those that respond to yellow colours, round shapes, brightness, moon, sky etc, etc.
The origin of these concepts is the world. To put it sloppily We grow concepts in our brains and our brains link them up. The nature of brains is interconnectedness. Brains make connections between responses to the world. Some of those connected responses are what we call ideas.
New connections are potentially new ideas, but more likely gibberish. A capacity to recognise combinations that are like other combinations that 'work' allows the gibberish to fade away and the stuff that works can be reinforced until it grows into permanence.
I doubt there is any fundamental difference at the neural network level between learning to control your arm and learning arithmetic or forming a concept like 'mother' or 'Sun'. They are different patterns that we learn to recognise because feedback indicates that they work in some way.