Review of Hawkins’s “On Intelligence”
Jonathan Hawkins, the creator of the PalmPilot and Graffiti handwriting software it uses, has written a book outlining a very interesting theory of what intelligence is. “On Intelligence” takes a non-behavioral (intelligence is not the same as intelligent behavior) and non-computational (intelligence is more than the ability to compute) approach to intelligence. Instead, he views intelligence as the ability to make predictions by taking stored memories and predicting future outcomes based on those memories. Hawkins suggests, uncontroversially, that this ability comes exclusively from the neocortex and, more controversially, that all operations we call intelligent can be reduced to the ability to make and adjust predictions.
Here is my review of the book, which I gave a four out of five stars on amazon.com
Jonathan Hawkins’s concern in “On Intelligence” is to outline a theory of what intelligence is that differs from ones floated around in various artificial intelligence (AI) circles. First, most theories of how to build “intelligent machines” focus exclusively on “intelligent behavior” without focus on the “thought” that must be behind it. (Think about Alan Turing’s test of an intelligent machine: if its behavior seems intelligent to humans, it must be intelligent. Purely behavioral.) Also, Hawkins is concerned that those few AI folks who have given thought to what intelligence is, apart from behavior, see intelligence as “ability to computer” and analogize it to a computer. But, Hawkins rightly notes, what we see as human intelligence -ability to synthesize disparate information, create novel solutions, apply old knowledge to new problems – is much more than computation.
Hawkins offers a very different theory to explain intelligence; intelligence, he writes, is the ability to predict outcomes. Hawkins goes a long way towards demonstrating how predicting outcomes is so much a part of what we do that we hardly even notice how much we do this. Every time we act, we predict (very subtly) the outcome of the action, and, because we predicted the outcome, are surprised when the plan goes awry. When I walk down the steps, I predict where each step will be so that I move my leg accordingly, and become alarmed when the step is not there. When I speak, I predict how the listener will interpret my words, and problem solve when the listener “misinterprets.”
As an educator, this book speaks to a very relevant part of my job. In order to enrich students’ intelligence, it is good to have an idea of what intelligence is so that we can teach the correct things via the correct methods. Hawkins theory that intelligence is the ability to predict and problem solve based on prediction seems a good description of what educators mean when we speak of intelligence. Intelligence tests do seem to test on tasks necessary for prediction. Some tests assess the ability to recognize patterns and predict future patterns (1, 3, 5, ?). Other tests look at the ability to recognize the relevant information from written passages (“Choose the option that best paraphrases the passage.”) while other tests assess the ability to openly make predictions (“What will happen next in the story?”).
If there is one flaw in Hawkins book, it is the discussion on creativity, which he reduces to the ability to make predictions. He makes a compelling case that creativity can be reduced this way in some instances: the mathematician solving a theorem. His case involving artistic creativity, and its blend of novelty and recycling, is a bit more tenuous. Despite Hawkins’s argument, I cannot see how writing a poem in any way involves making a prediction (unless the prediction is about how the poem will be received). My guess is that some aspects of creativity involve intelligence (what words rhyme with ___? Will readers understand this metaphor?”) Other areas (what metaphor can I create for this experience? Does an “open” or “closed” vowel sound better to end this line?) cannot be reduced to basic prediction. (I also have a problem with Hawkins’s discussion of what consciousness is, which I think is question-begging, but I won’t elaborate here. Read it for yourself and see what you think.
Anyhow, Hawkins book and theory is very eye-opening and, in my view, more “on the money” than the computational or behavioral model of intelligence. This is especially so in light of the fact that Hawkins’s used this theory of intelligence as ability to predict and learn via memory storage to create the PalmPilot and the Graffiti handwriting software it uses. Much recommended for anyone interested in reading a different theory of intelligence – especially educators. [For those interested in a book that compliments this one, read What Intelligence Tests Miss: The Psychology of Rational Thought, which broadens our standard definition of intelligence to include the ability to reason and solve problems based on stored information, much like Hawkins does.