In the chapter 10 of his book A new Kind of science (2012), Stephen Wolfram gives an idea, based in his principle of computational equivalence, how he thinks that the perception, storage and recovering of information in mind can be explained computationally.
Code available here
Of course, this explanation -as Wolfram himself says- "does not means that this is how really minds works", but it is and idea of how can be explained some basic mind tasks.
He starts saying that our senses have a kind of filters of information in such a way that not all information around us reach our minds but only relevant details.
Information that reach our insides is coded in some way (he never gave technical details), and such codifications works at the same time as address where information must be hold.
Then, since only relevant aspects of information are codified, an efficient algorithm of searching and recovering information can be implemented in order to bring relevant information to react or to give an answer back.
The most important think in the Wolfram's proposal is perhaps his claim that it is not necessary to save every detail of information that we receive from environment, but only relevant aspects useful to reconstruct whole ideas or information.
Here, in this project I found a way to implemented these ideas in order to develop an algorithm to filter (code/abstraction), save (in efficient manner where similar information remains together) and recover information (by similarity).
When a search of information take place, then, In the same spirit of the Wolfram`s idea, from relevant aspects it is capable to construct the whole original information.
In this picture can be seen in the up left corner original information. In the down left corner the filtered information using Boltzmann networks. In the up right corner it is the filtered information, this is the resumed version of the original one.
In the down right information it is the step of recovering information from the resumed version of it.
I introduced a paper on critics to this work in IACAP 2014.