# From Deep Learning to Rational Machines
## Metadata
- Author: [[Cameron J. Buckner;]]
- Full Title: From Deep Learning to Rational Machines
- Category: #books
- Summary: insert summary
- My notes:
- Summary: The text discusses the limitations of current deep learning models in understanding complex cognitive functions like metacognition and social cognition. It emphasizes the role of imagination and attention in rational thought and suggests that integrating language modules could enhance these models. Additionally, it highlights how humans and animals use the same neural structures for both social and spatial reasoning, indicating the versatility of cognitive processes.
- Source File: From Deep Learning to Rational Machines - Cameron J. Buckner;.txt
## LLM Chats
## NotebookLM
## LLM Audio
## Highlights
> In this section, I propose a framework for understanding abstraction that bears some affinity to all four forms just discussed; and in the subsequent section, I show how DCNNs highlight the aspects of the mammalian brain that perform these forms of abstraction in a way that is consistent with the Transformation Principle, cashing out a significant promissory note for empiricism. ([View Highlight](https://read.readwise.io/read/01jb4mgkbdtnb2f96ea8e614t4))
> As a case in point, one particular move made by AlphaGo against Lee—move 37 of Game 2—was so unorthodox and unexpected that even the DeepMind employee tasked with recording the move assumed it must have been a bug. On the contrary, commentators now describe this move as exceptionally brilliant. Supported by no conventional strategy or theory, AlphaGo on move 37 opted to place a Go stone in an area of open space on the right side of the board that seemed to provide little advantage to nearby pieces’ influence or stability, forgoing several more obvious moves that would have shored up its position ([Fig. 2.4](private://read/01jb4k14fesb5g4mk0z27jyn6f/#oso-9780197653302-chapter-2-figureGroup-7)). The match’s commentators at the time (one also a 9-dan grandmaster of the game) were dumbstruck, and Lee himself was so troubled by the move that he immediately left the room to compose himself and could not formulate a response for nearly fifteen minutes. Yet after ten or twenty more turns, this move was revealed as a crucial turning point in the match, which AlphaGo won handily. ([View Highlight](https://read.readwise.io/read/01jdj1fnc0y8cjr4mqmndg52h2))