In the space of all useful algorithms, there are algorithms that human can repeat in their minds and there are algorithms that are the building blocks for their own thoughts. There is no law that guarantees that they are the same.
Said differently, algorithms that we show that can solve a problem or usually not the same algorithm a human may have used to solve the same problem. But because we have are educated, we mistake the algorithm we learned to use to be the same algorithm that we intrinsically have.
The class of algorithms that serve as the building blocks of human cognition is the class of algorithms that can be learned from interaction with the world and without explicit instruction. But learning itself is an algorithm, so the human algorithm must be a learning algorithm.
The problem with most algorithms are that they are not learning algorithms. They are algorithms that can be learned but that's not the same thing.
There are very few learning algorithms that we know about. We know of evolution. We know of machine learning and deep learning. Cultural evolution is also a learning algorithm. What are universal about these methods? What are the building blocks of learning algorithms?
We cannot even begin to understand a Theory of Intelligence if we cannot at least identify the universal components of learning algorithms.
What is indeed very intriguing is that we have yet to discover the learning algorithm of the human brain. The closest depiction of this algorithm is the notion of analogy making.
This is the purpose of this new book: "Empathy Emergence". To add more content to the understanding of the missing link of general intelligence. 
@threadreaderapp unroll for future reference.
