#Book #AI #Psychology/Cognition #Neuroscience #2021/6 *June 7, 2021* ![[on-intelligence.png]] [Amazon](https://www.amazon.com/Intelligence-Understanding-Creation-Intelligent-Machines-ebook/dp/B003J4VE5Y/ref=sr_1_1?dchild=1&keywords=on+intelligence&qid=1623116691&sr=8-1) ### Summary *On Intelligence* offers a theory of how the brain couches intelligence called the [[Memory-Prediction Framework]]. The gist of this theory is that since the cortex seems to have a uniform tissue, it is reasonable to expect that the cortex processes information in the same way throughout, no matter whether the regions of the cortex are processing auditory, visual, touch, language, or motor signals. Our memories are built into the connections between neurons in the cortex, and we use those connections to make predictions because signals in the cortex can be fed back through memory circuits to predict what will happen next in a sequence of events. This is an example of [[Recurrent Networks]]. For example, if you hear someone say 'I'm going to eat this banana', there are connections in your cortex from the language centers to the visual centers that allow you to expect the person to eat a banana, and even connections to your sensory regions that allow you to predict what that banana will taste like, and what it will feel like to eat it. The book also contains some discussion of more philosophical topics such as consciousness and creativity, but this is less interesting than Hawkins's theory of intelligence. For example, his theory of consciousness basically boils down to "consciousness is what it feels like to have \[insert information-processing mechanism here\]", and his discussion of creativity basically rehashes the psychology of creativity. I definitely recommend reading these sections if you aren't familiar with these topics, but otherwise, you may find them underwhelming. The same is true of the section dealing with predictions of what it will take to build intelligent machines. I think the value of this book for me was mostly in establishing a strong sense of what 'intelligence' is, as well as tying it in to what we know about how the brain works. It also provides some good jumping-off points into the neuroscience literature that I plan to follow.